[ARM] Add MVE vector load/store instructions.

[simon_tatham]

Summary

This adds the full set of vector memory access instructions. It
includes contiguous loads/stores, with an ordinary addressing mode
such as [r0,#offset] (plus writeback variants); gather loads and
scatter stores with a scalar base address register and a vector of
offsets from it (written [r0,q1] or similar); and gather/scatters with
a vector of base addresses (written [q0,#offset], again with
writeback). Additionally, some of the loads can widen each loaded
value into a larger vector lane, and the corresponding stores narrow
them again. Finally, there's the VLD2 / VLD4 family, which distributes
2 or 4 vectors' worth of memory data across the lanes of the same
number of registers but in a transposed order.

To implement these, we also have to add the addressing modes they
need, and the register list operands used by VLD2/VST4. Also, in
AsmParser, the isMem query function now has subqueries isGPRMem
and isMVEMem, according to which kind of base register is used by a
given memory access operand.

[simon_tatham]

Removed some code duplication in ARMMCCodeEmitter.cpp, arising from review of D62667.

[samparker]

llvm/lib/Target/ARM/ARMInstrMVE.td

3940

These descriptions would be far easier to read if the load and store were separated. That would also make them easier to use when we want to apply isel patterns. I can't say that I've seen this syntax used in this way before... probably for good reason.

4015

Looks like these mayLoad/Store flags have been mixed up.

4066

Again, I'm assuming that this incorrect, IndexModeNone instead?

4151

and here.

4188

What is f8?

4302

Should this 'ins' be passed as an argument? Looks like all the instructions pass it.

4412

Since we're already passing 'shift', it would be cleaner to construct the addrOps within the multiclass definitions.

llvm/lib/Target/ARM/ARMInstrThumb2.td

323

Could this stuff be moved into the MVE specific td file?

llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp

1755

Please extract out the shared logic into a helper.

2548

either use cast if you're sure, otherwise we need to ensure CE isn't nullptr.

2977

Do we really need all these functions that do exactly the same thing?!

4690–4691

NoLanes and AllLanes can be folded.

7468

Please use a helper to avoid the copy-paste.

llvm/lib/Target/ARM/Disassembler/ARMDisassembler.cpp

6333

This functions share a lot of common functionality, please refactor.

[simon_tatham]

Remastered patch to apply cleanly against current trunk.

[simon_tatham]

(I'm still working on the rest of your review comments)

llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp

2548

That hardly seems fair – all the other functions on both sides of this hunk are written in the same style! But I'll add an assertion to just these ones if you like :-)

2977

I think all the names have to exist, because the Tablegen-generated code will want to call them all by concatenating the innermost operand name (Imm7Shift1 or similar) into the middle of a standard function-name template. But I can turn them all into trivial wrappers on a single function body.

[ostannard]

llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp

2977

You can override the function name used by setting the RenderMethod field of the AsmOperandClass, so these could all use the same function.

[simon_tatham]

I've decided this patch is too large to manage all in one go. Also, the interleaving family of loads (VLD20 and friends) share essentially no infrastructure with the VLDR family, so that seems like a natural place to split the patch in two.

Hence, I've just opened D63650, containing just the VLD20 family, hopefully with all comments on this patch taken into account. I'll keep this review open and use it for the remainder of the loads and stores, but I haven't finished reworking it yet.

[simon_tatham]

Reworked the remaining loads and stores to address review comments, implement consistent naming of instruction ids, and tidy up the Tablegen so that it's hopefully halfway readable.

I think this is still the most complicated subset of instructions in the whole of MVE, but I'm running out of ways I can make it simpler.

[simon_tatham]

Minor revisions to this patch: changed a couple of legacy t2rGPR into rGPR (after the former was withdrawn during code review of D63650). Also added a small knock-on fix in checkTargetMatchPredicate, preventing an assertion failure in a check that was specific to rGPR.

[ostannard]

llvm/lib/Target/ARM/ARMInstrMVE.td

161

I think this can be written like this, without the defs above:

let ParserMatchClass = TMemImm7ShiftOffsetAsmOperand<shift>;

3611

Why is this needed, could we just define instances of MVE_VLDRSTR_rq directly below?

llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp

5812

Missing space before equals

7487

I don't see any tests for the QmIsPointer version of this error.

llvm/lib/Target/ARM/Disassembler/ARMDisassembler.cpp

4246

This doesn't need to be on it's own line.

llvm/lib/Target/ARM/Thumb2InstrInfo.cpp

617–622

I think it would be clearer to define the values for all three AddrModes in the switch, rather than using these defaults for one of them.

llvm/test/MC/ARM/mve-load-store.s

3

None of these instructions differ with mve.fp (unless I missed some), so there's no need to duplicate the check lines. We should also be checking the error messages in the no-fp case.

[simon_tatham]

llvm/lib/Target/ARM/ARMInstrMVE.td

3611

The idea was partly to abstract away all the parameters to MVE_VLDRSTR_rq that are common between the byte variants, and mostly to make all the defms look similar enough that the reader can focus on only the necessary differences.

Perhaps a one-element defm was silly and I should turn it into an ordinary subclass, though.

[simon_tatham]

Addressed all review comments, I think.

[ostannard]

LGTM

[kcc]

This new test is currently failing on the ubsan build.
Could you please take a look?
I don't yet know is the breaking change.

http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds/33133/steps/check-llvm%20ubsan/logs/stdio

FAIL: LLVM :: MC/Disassembler/ARM/mve-load-store.txt (22713 of 32124)

• TEST 'LLVM :: MC/Disassembler/ARM/mve-load-store.txt' FAILED ****

Script:

: 'RUN: at line 1'; /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/bin/llvm-mc -disassemble -triple=thumbv8.1m.main-none-eabi -mattr=+mve.fp,+fp64 -show-encoding /b/sanitizer-x86_64-linux-fast/build/llvm/test/MC/Disassembler/ARM/mve-load-store.txt 2> /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/test/MC/Disassembler/ARM/Output/mve-load-store.txt.tmp | /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/bin/FileCheck /b/sanitizer-x86_64-linux-fast/build/llvm/test/MC/Disassembler/ARM/mve-load-store.txt
: 'RUN: at line 2'; /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/bin/FileCheck --check-prefix=ERROR < /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/test/MC/Disassembler/ARM/Output/mve-load-store.txt.tmp /b/sanitizer-x86_64-linux-fast/build/llvm/test/MC/Disassembler/ARM/mve-load-store.txt
: 'RUN: at line 3'; not /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/bin/llvm-mc -disassemble -triple=thumbv8.1m.main-none-eabi -show-encoding /b/sanitizer-x86_64-linux-fast/build/llvm/test/MC/Disassembler/ARM/mve-load-store.txt &> /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/test/MC/Disassembler/ARM/Output/mve-load-store.txt.tmp

: 'RUN: at line 4'; /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/bin/FileCheck --check-prefix=CHECK-NOMVE < /b/sanitizer-x86_64-linux-fast/build/llvm_build_ubsan/test/MC/Disassembler/ARM/Output/mve-load-store.txt.tmp /b/sanitizer-x86_64-linux-fast/build/llvm/test/MC/Disassembler/ARM/mve-load-store.txt

Exit Code: 1

Command Output (stderr):

/b/sanitizer-x86_64-linux-fast/build/llvm/test/MC/Disassembler/ARM/mve-load-store.txt:269:10: error: CHECK: expected string not found in input

CHECK: vldrb.u32 q0, [r0] @ encoding: [0x90,0xfd,0x00,0x0f]

^

<stdin>:68:2: note: scanning from here
vldr
^

[kcc]

Actually, the test started failing in ubsan mode right after this commit. Please fix ASAP. Reproduce: build with -DLLVM_USE_SANITIZER=Undefined, then "ninja check-llvm"

[simon_tatham]

Sorry about that. rL364635 should fix it.

llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp

if (Opcode == ARM::ADDri) {
     case ARMII::AddrMode5FP16:
       ImmIdx = FrameRegIdx+1;
       InstrOffs = ARM_AM::getAM5Offset(MI.getOperand(ImmIdx).getImm());
       if (ARM_AM::getAM5Op(MI.getOperand(ImmIdx).getImm()) == ARM_AM::sub)
         InstrOffs *= -1;
       NumBits = 8;
       Scale = 2;
       break;
+    case ARMII::AddrModeT2_i7:
+    case ARMII::AddrModeT2_i7s2:
     case ARMII::AddrModeT2_i7s4:
       ImmIdx = FrameRegIdx+1;
       InstrOffs = MI.getOperand(ImmIdx).getImm();
       NumBits = 7;
-      Scale = 4;
+      Scale = (AddrMode == ARMII::AddrModeT2_i7s2 ? 2 :
+               AddrMode == ARMII::AddrModeT2_i7s4 ? 4 : 1);
       break;
     default:
       llvm_unreachable("Unsupported addressing mode!");
     }
 
     Offset += InstrOffs * Scale;
     assert((Offset & (Scale-1)) == 0 && "Can't encode this offset!");
     if (Offset < 0) {

llvm/lib/Target/ARM/ARMInstrFormats.td

 def AddrModeT2_i8   : AddrMode<12>;
 def AddrModeT2_so   : AddrMode<13>;
 def AddrModeT2_pc   : AddrMode<14>;
 def AddrModeT2_i8s4 : AddrMode<15>;
 def AddrMode_i12    : AddrMode<16>;
 def AddrMode5FP16   : AddrMode<17>;
 def AddrModeT2_ldrex : AddrMode<18>;
 def AddrModeT2_i7s4 : AddrMode<19>;
+def AddrModeT2_i7s2 : AddrMode<20>;
+def AddrModeT2_i7   : AddrMode<21>;
 
 // Load / store index mode.
 class IndexMode<bits<2> val> {
   bits<2> Value = val;
 }
 def IndexModeNone : IndexMode<0>;
 def IndexModePre  : IndexMode<1>;
 def IndexModePost : IndexMode<2>;

llvm/lib/Target/ARM/ARMInstrInfo.td

 def addr_offset_none : MemOperand,
                        ComplexPattern<i32, 1, "SelectAddrOffsetNone", []> {
   let PrintMethod = "printAddrMode7Operand";
   let DecoderMethod = "DecodeAddrMode7Operand";
   let ParserMatchClass = MemNoOffsetAsmOperand;
   let MIOperandInfo = (ops GPR:$base);
 }
 
+// t_addr_offset_none := reg [r0-r7]
+def MemNoOffsetTAsmOperand : AsmOperandClass { let Name = "MemNoOffsetT"; }
+def t_addr_offset_none : MemOperand {
+  let PrintMethod = "printAddrMode7Operand";
+  let DecoderMethod = "DecodetGPRRegisterClass";
+  let ParserMatchClass = MemNoOffsetTAsmOperand;
+  let MIOperandInfo = (ops tGPR:$base);
+}
+
 def nohash_imm : Operand<i32> {
   let PrintMethod = "printNoHashImmediate";
 }
 
 def CoprocNumAsmOperand : AsmOperandClass {
   let Name = "CoprocNum";
   let ParserMethod = "parseCoprocNumOperand";
 }

llvm/lib/Target/ARM/ARMInstrMVE.td

def VecList4QAsmOperand : AsmOperandClass {
   let ParserMethod = "parseVectorList";
   let RenderMethod = "addMVEVecListOperands";
 }
 
 def VecList4Q : RegisterOperand<QQQQPR, "printMVEVectorListFourQ"> {
   let ParserMatchClass = VecList4QAsmOperand;
 }
 
+// taddrmode_imm7  := reg[r0-r7] +/- (imm7 << shift)
+class TMemImm7ShiftOffsetAsmOperand<int shift> : AsmOperandClass {
+  let Name = "TMemImm7Shift"#shift#"Offset";
+  let PredicateMethod = "isMemImm7ShiftedOffset<"#shift#",ARM::tGPRRegClassID>";
+  let RenderMethod = "addMemImmOffsetOperands";
+}
+
+def TMemImm7Shift0OffsetAsmOperand : TMemImm7ShiftOffsetAsmOperand<0>;
+def TMemImm7Shift1OffsetAsmOperand : TMemImm7ShiftOffsetAsmOperand<1>;
+
+class taddrmode_imm7<int shift> : MemOperand {
+  let ParserMatchClass =
+   !cast<AsmOperandClass>("TMemImm7Shift"#shift#"OffsetAsmOperand");
+  // They are printed the same way as the T2 imm8 version
+  let PrintMethod = "printT2AddrModeImm8Operand<false>";
+  // This can also be the same as the T2 version.
+  let EncoderMethod = "getT2AddrModeImmOpValue<7,"#shift#">";
+  let DecoderMethod = "DecodeTAddrModeImm7<"#shift#">";
+  let MIOperandInfo = (ops tGPR:$base, i32imm:$offsimm);
+}
+
+// t2addrmode_imm7  := reg +/- (imm7)
+class MemImm7ShiftOffsetAsmOperand<int shift> : AsmOperandClass {
+  let Name = "MemImm7Shift"#shift#"Offset";
+  let PredicateMethod = "isMemImm7ShiftedOffset<" # shift #
+                        ",ARM::GPRnopcRegClassID>";
+  let RenderMethod = "addMemImmOffsetOperands";
+}
+
+def MemImm7Shift0OffsetAsmOperand : MemImm7ShiftOffsetAsmOperand<0>;
+def MemImm7Shift1OffsetAsmOperand : MemImm7ShiftOffsetAsmOperand<1>;
+def MemImm7Shift2OffsetAsmOperand : MemImm7ShiftOffsetAsmOperand<2>;
+class T2AddrMode_Imm7<int shift> : MemOperand,
+      ComplexPattern<i32, 2, "SelectT2AddrModeImm7<"#shift#">", []> {
+  let EncoderMethod = "getT2AddrModeImmOpValue<7,"#shift#">";
+  let DecoderMethod = "DecodeT2AddrModeImm7<"#shift#", 0>";
+  let ParserMatchClass =
+    !cast<AsmOperandClass>("MemImm7Shift"#shift#"OffsetAsmOperand");
+  let MIOperandInfo = (ops GPRnopc:$base, i32imm:$offsimm);
+}
+
+class t2addrmode_imm7<int shift> : T2AddrMode_Imm7<shift> {
+  // They are printed the same way as the imm8 version
+  let PrintMethod = "printT2AddrModeImm8Operand<false>";
+}
+
+class MemImm7ShiftOffsetWBAsmOperand<int shift> : AsmOperandClass {
+  let Name = "MemImm7Shift"#shift#"OffsetWB";
+  let PredicateMethod = "isMemImm7ShiftedOffset<" # shift #
+                        ",ARM::rGPRRegClassID>";
+  let RenderMethod = "addMemImmOffsetOperands";
+}
+
+def MemImm7Shift0OffsetWBAsmOperand : MemImm7ShiftOffsetWBAsmOperand<0>;
+def MemImm7Shift1OffsetWBAsmOperand : MemImm7ShiftOffsetWBAsmOperand<1>;
+def MemImm7Shift2OffsetWBAsmOperand : MemImm7ShiftOffsetWBAsmOperand<2>;
+
+class t2addrmode_imm7_pre<int shift> : T2AddrMode_Imm7<shift> {
+  // They are printed the same way as the imm8 version
+  let PrintMethod = "printT2AddrModeImm8Operand<true>";
+  let ParserMatchClass =
+    !cast<AsmOperandClass>("MemImm7Shift"#shift#"OffsetWBAsmOperand");
+  let DecoderMethod = "DecodeT2AddrModeImm7<"#shift#", 1>";
+  let MIOperandInfo = (ops rGPR:$base, i32imm:$offsim);
+}
+
+class t2am_imm7shiftOffsetAsmOperand<int shift>
+  : AsmOperandClass { let Name = "Imm7Shift"#shift; }
+def t2am_imm7shift0OffsetAsmOperand : t2am_imm7shiftOffsetAsmOperand<0>;
+def t2am_imm7shift1OffsetAsmOperand : t2am_imm7shiftOffsetAsmOperand<1>;
+def t2am_imm7shift2OffsetAsmOperand : t2am_imm7shiftOffsetAsmOperand<2>;
+
+class t2am_imm7_offset<int shift> : MemOperand {
+  // They are printed the same way as the imm8 version
+  let PrintMethod = "printT2AddrModeImm8OffsetOperand";
+  let ParserMatchClass =
+    !cast<AsmOperandClass>("t2am_imm7shift"#shift#"OffsetAsmOperand");
+  let EncoderMethod = "getT2ScaledImmOpValue<7,"#shift#">";
+  let DecoderMethod = "DecodeT2Imm7<"#shift#">";
+}
+
+// Operands for gather/scatter loads of the form [Rbase, Qoffsets]
+class MemRegRQOffsetAsmOperand<int shift> : AsmOperandClass {
+  let Name = "MemRegRQS"#shift#"Offset";
+  let PredicateMethod = "isMemRegRQOffset<"#shift#">";
+  let RenderMethod = "addMemRegRQOffsetOperands";
+}
+
+def MemRegRQS0OffsetAsmOperand : MemRegRQOffsetAsmOperand<0>;
+def MemRegRQS1OffsetAsmOperand : MemRegRQOffsetAsmOperand<1>;
+def MemRegRQS2OffsetAsmOperand : MemRegRQOffsetAsmOperand<2>;
+def MemRegRQS3OffsetAsmOperand : MemRegRQOffsetAsmOperand<3>;
+
+// mve_addr_rq_shift  := reg + vreg{ << UXTW #shift}
+class mve_addr_rq_shift<int shift> : MemOperand {
+  let EncoderMethod = "getMveAddrModeRQOpValue";
+  let PrintMethod = "printMveAddrModeRQOperand<"#shift#">";
+  let ParserMatchClass =
+    !cast<AsmOperandClass>("MemRegRQS"#shift#"OffsetAsmOperand");
+  let DecoderMethod = "DecodeMveAddrModeRQ";
+  let MIOperandInfo = (ops GPRnopc:$base, MQPR:$offsreg);
+}
+
+class MemRegQOffsetAsmOperand<int shift> : AsmOperandClass {
+  let Name = "MemRegQS"#shift#"Offset";
+  let PredicateMethod = "isMemRegQOffset<"#shift#">";
+  let RenderMethod = "addMemImmOffsetOperands";
+}
+
+def MemRegQS2OffsetAsmOperand : MemRegQOffsetAsmOperand<2>;
+def MemRegQS3OffsetAsmOperand : MemRegQOffsetAsmOperand<3>;
+
+// mve_addr_q_shift  := vreg {+ #imm7s2/4}
+class mve_addr_q_shift<int shift> : MemOperand {
+  let EncoderMethod = "getMveAddrModeQOpValue<"#shift#">";
+  // Can be printed same way as other reg + imm operands
+  let PrintMethod = "printT2AddrModeImm8Operand<false>";
+  let ParserMatchClass =
+    !cast<AsmOperandClass>("MemRegQS"#shift#"OffsetAsmOperand");
+  let DecoderMethod = "DecodeMveAddrModeQ<"#shift#">";
+  let MIOperandInfo = (ops MQPR:$base, i32imm:$imm);
+}
+
+// --------- Start of base classes for the instructions themselves
+
 class MVE_MI<dag oops, dag iops, InstrItinClass itin, string asm,
              string ops, string cstr, list<dag> pattern>
   : Thumb2XI<oops, iops, AddrModeNone, 4, itin, !strconcat(asm, "\t", ops), cstr,
              pattern>,
     Requires<[HasMVEInt]> {
   let D = MVEDomain;
   let DecoderNamespace = "MVE";
 }
foreach wb = [MVE_vldst24_writeback<
 
   def "MVE_VST" # n.nvecs # stage # "_" # s.lanesize # wb.id_suffix
     : MVE_vst24_base<n, stage, s.sizebits, wb,
                      "vst" # n.nvecs # stage # "." # s.lanesize>;
 }
 
 // end of MVE interleaving load/store
 
+// start of MVE predicable load/store
+
+// A parameter class for the direction of transfer.
+class MVE_ldst_direction<bit b, dag Oo, dag Io, string c=""> {
+  bit load = b;
+  dag Oops = Oo;
+  dag Iops = Io;
+  string cstr = c;
+}
+def MVE_ld: MVE_ldst_direction<1, (outs MQPR:$Qd), (ins), ",@earlyclobber $Qd">;
+def MVE_st: MVE_ldst_direction<0, (outs), (ins MQPR:$Qd)>;
+
+// A parameter class for the size of memory access in a load.
+class MVE_memsz<bits<2> e, int s, AddrMode m, string mn, list<string> types> {
+  bits<2> encoding = e;         // opcode bit(s) for encoding
+  int shift = s;                // shift applied to immediate load offset
+  AddrMode AM = m;
+
+  // For instruction aliases: define the complete list of type
+  // suffixes at this size, and the canonical ones for loads and
+  // stores.
+  string MnemonicLetter = mn;
+  int TypeBits = !shl(8, s);
+  string CanonLoadSuffix = ".u" # TypeBits;
+  string CanonStoreSuffix = "." # TypeBits;
+  list<string> suffixes = !foreach(letter, types, "." # letter # TypeBits);
+}
+
+// Instances of MVE_memsz.
+//
+// (memD doesn't need an AddrMode, because those are only for
+// contiguous loads, and memD is only used by gather/scatters.)
+def MVE_memB: MVE_memsz<0b00, 0, AddrModeT2_i7,   "b", ["", "u", "s"]>;
+def MVE_memH: MVE_memsz<0b01, 1, AddrModeT2_i7s2, "h", ["", "u", "s", "f"]>;
+def MVE_memW: MVE_memsz<0b10, 2, AddrModeT2_i7s4, "w", ["", "u", "s", "f"]>;
+def MVE_memD: MVE_memsz<0b11, 3, ?,               "d", ["", "u", "s", "f"]>;
+
+// This is the base class for all the MVE loads and stores other than
+// the interleaving ones. All the non-interleaving loads/stores share
+// the characteristic that they operate on just one vector register,
+// so they are VPT-predicable.
+//
+// The predication operand is vpred_n, for both loads and stores. For
+// store instructions, the reason is obvious: if there is no output
+// register, there can't be a need for an input parameter giving the
+// output register's previous value. Load instructions also don't need
+// that input parameter, because unlike MVE data processing
+// instructions, predicated loads are defined to set the inactive
+// lanes of the output register to zero, instead of preserving their
+// input values.
+class MVE_VLDRSTR_base<MVE_ldst_direction dir, bit U, bit P, bit W, bit opc,
+                       dag oops, dag iops, string asm, string suffix,
+                       string ops, string cstr, list<dag> pattern=[]>
+ : MVE_p<oops, iops, NoItinerary, asm, suffix, ops, vpred_n, cstr, pattern> {
+  bits<3> Qd;
+
+  let Inst{28} = U;
+  let Inst{25} = 0b0;
+  let Inst{24} = P;
+  let Inst{22} = 0b0;
+  let Inst{21} = W;
+  let Inst{20} = dir.load;
+  let Inst{15-13} = Qd{2-0};
+  let Inst{12} = opc;
+  let Inst{11-9} = 0b111;
+
+  let mayLoad = dir.load;
+  let mayStore = !eq(dir.load,0);
+}
+
+// Contiguous load and store instructions. These come in two main
+// categories: same-size loads/stores in which 128 bits of vector
+// register is transferred to or from 128 bits of memory in the most
+// obvious way, and widening loads / narrowing stores, in which the
+// size of memory accessed is less than the size of a vector register,
+// so the load instructions sign- or zero-extend each memory value
+// into a wider vector lane, and the store instructions truncate
+// correspondingly.
+//
+// The instruction mnemonics for these two classes look reasonably
+// similar, but the actual encodings are different enough to need two
+// separate base classes.
+
+// Contiguous, same size
+class MVE_VLDRSTR_cs<MVE_ldst_direction dir, MVE_memsz memsz, bit P, bit W,
+                     dag oops, dag iops, string asm, string suffix,
+                     IndexMode im, string ops, string cstr>
+  : MVE_VLDRSTR_base<dir, 0, P, W, 1, oops, iops, asm, suffix, ops, cstr> {
+  bits<12> addr;
+  let Inst{23} = addr{7};
+  let Inst{19-16} = addr{11-8};
+  let Inst{8-7} = memsz.encoding;
+  let Inst{6-0} = addr{6-0};
+}
+
+// Contiguous, widening/narrowing
+class MVE_VLDRSTR_cw<MVE_ldst_direction dir, MVE_memsz memsz, bit U,
+                     bit P, bit W, bits<2> size, dag oops, dag iops,
+                     string asm, string suffix, IndexMode im,
+                     string ops, string cstr>
+  : MVE_VLDRSTR_base<dir, U, P, W, 0, oops, iops, asm, suffix, ops, cstr> {
+  bits<11> addr;
+  let Inst{23} = addr{7};
+  let Inst{19} = memsz.encoding{0}; // enough to tell 16- from 32-bit
+  let Inst{18-16} = addr{10-8};
+  let Inst{8-7} = size;
+  let Inst{6-0} = addr{6-0};
+
+  let IM = im;
+}
+
+// Multiclass wrapper on each of the _cw and _cs base classes, to
+// generate three writeback modes (none, preindex, postindex).
+
+multiclass MVE_VLDRSTR_cw_m<MVE_ldst_direction dir, MVE_memsz memsz,
+                            string asm, string suffix, bit U, bits<2> size> {
+  let AM = memsz.AM in {
+    def "" : MVE_VLDRSTR_cw<
+        dir, memsz, U, 1, 0, size,
+        dir.Oops, !con(dir.Iops, (ins taddrmode_imm7<memsz.shift>:$addr)),
+        asm, suffix, IndexModeNone, "$Qd, $addr", "">;
+
+    def _pre : MVE_VLDRSTR_cw<
+        dir, memsz, U, 1, 1, size,
+        !con((outs tGPR:$wb), dir.Oops),
+        !con(dir.Iops, (ins taddrmode_imm7<memsz.shift>:$addr)),
+        asm, suffix, IndexModePre, "$Qd, $addr!", "$addr.base = $wb"> {
+      let DecoderMethod = "DecodeMVE_MEM_1_pre<"#memsz.shift#">";
+    }
+
+    def _post : MVE_VLDRSTR_cw<
+        dir, memsz, U, 0, 1, size,
+        !con((outs tGPR:$wb), dir.Oops),
+        !con(dir.Iops, (ins t_addr_offset_none:$Rn,
+                            t2am_imm7_offset<memsz.shift>:$addr)),
+        asm, suffix, IndexModePost, "$Qd, $Rn$addr", "$Rn.base = $wb"> {
+      bits<4> Rn;
+      let Inst{18-16} = Rn{2-0};
+    }
+  }
+}
+
+multiclass MVE_VLDRSTR_cs_m<MVE_ldst_direction dir, MVE_memsz memsz,
+                            string asm, string suffix> {
+  let AM = memsz.AM in {
+    def "" : MVE_VLDRSTR_cs<
+        dir, memsz, 1, 0,
+        dir.Oops, !con(dir.Iops, (ins t2addrmode_imm7<memsz.shift>:$addr)),
+        asm, suffix, IndexModeNone, "$Qd, $addr", "">;
+
+    def _pre : MVE_VLDRSTR_cs<
+        dir, memsz, 1, 1,
+        !con((outs t2rGPR:$wb), dir.Oops),
+        !con(dir.Iops, (ins t2addrmode_imm7_pre<memsz.shift>:$addr)),
+        asm, suffix, IndexModePre, "$Qd, $addr!", "$addr.base = $wb"> {
+      let DecoderMethod = "DecodeMVE_MEM_2_pre<"#memsz.shift#">";
+    }
+
+    def _post : MVE_VLDRSTR_cs<
+        dir, memsz, 0, 1,
+        !con((outs t2rGPR:$wb), dir.Oops),
+        // We need an !if here to select the base register class,
+        // because it's legal to write back to SP in a load of this
+        // type, but not in a store.
+        !con(dir.Iops, (ins !if(dir.load, t2_addr_offset_none,
+                                          t2_nosp_addr_offset_none):$Rn,
+                            t2am_imm7_offset<memsz.shift>:$addr)),
+        asm, suffix, IndexModePost, "$Qd, $Rn$addr", "$Rn.base = $wb"> {
+      bits<4> Rn;
+      let Inst{19-16} = Rn{3-0};
+    }
+  }
+}
+
+// Now actually declare all the contiguous load/stores, via those
+// multiclasses. The instruction ids coming out of this are the bare
+// names shown in the defm, with _pre or _post appended for writeback,
+// e.g. MVE_VLDRBS16, MVE_VSTRB16_pre, MVE_VSTRHU16_post.
+
+defm MVE_VLDRBS16: MVE_VLDRSTR_cw_m<MVE_ld, MVE_memB, "vldrb", "s16", 0, 0b01>;
+defm MVE_VLDRBS32: MVE_VLDRSTR_cw_m<MVE_ld, MVE_memB, "vldrb", "s32", 0, 0b10>;
+defm MVE_VLDRBU16: MVE_VLDRSTR_cw_m<MVE_ld, MVE_memB, "vldrb", "u16", 1, 0b01>;
+defm MVE_VLDRBU32: MVE_VLDRSTR_cw_m<MVE_ld, MVE_memB, "vldrb", "u32", 1, 0b10>;
+defm MVE_VLDRHS32: MVE_VLDRSTR_cw_m<MVE_ld, MVE_memH, "vldrh", "s32", 0, 0b10>;
+defm MVE_VLDRHU32: MVE_VLDRSTR_cw_m<MVE_ld, MVE_memH, "vldrh", "u32", 1, 0b10>;
+
+defm MVE_VLDRBU8:  MVE_VLDRSTR_cs_m<MVE_ld, MVE_memB, "vldrb", "u8">;
+defm MVE_VLDRHU16: MVE_VLDRSTR_cs_m<MVE_ld, MVE_memH, "vldrh", "u16">;
+defm MVE_VLDRWU32: MVE_VLDRSTR_cs_m<MVE_ld, MVE_memW, "vldrw", "u32">;
+
+defm MVE_VSTRB16:  MVE_VLDRSTR_cw_m<MVE_st, MVE_memB, "vstrb", "16",  0, 0b01>;
+defm MVE_VSTRB32:  MVE_VLDRSTR_cw_m<MVE_st, MVE_memB, "vstrb", "32",  0, 0b10>;
+defm MVE_VSTRH32:  MVE_VLDRSTR_cw_m<MVE_st, MVE_memH, "vstrh", "32",  0, 0b10>;
+
+defm MVE_VSTRBU8 : MVE_VLDRSTR_cs_m<MVE_st, MVE_memB, "vstrb", "8">;
+defm MVE_VSTRHU16: MVE_VLDRSTR_cs_m<MVE_st, MVE_memH, "vstrh", "16">;
+defm MVE_VSTRWU32: MVE_VLDRSTR_cs_m<MVE_st, MVE_memW, "vstrw", "32">;
+
+// Gather loads / scatter stores whose address operand is of the form
+// [Rn,Qm], i.e. a single GPR as the common base address, plus a
+// vector of offset from it. ('Load/store this sequence of elements of
+// the same array.')
+//
+// Like the contiguous family, these loads and stores can widen the
+// loaded values / truncate the stored ones, or they can just
+// load/store the same size of memory and vector lane. But unlike the
+// contiguous family, there's no particular difference in encoding
+// between those two cases.
+//
+// This family also comes with the option to scale the offset values
+// in Qm by the size of the loaded memory (i.e. to treat them as array
+// indices), or not to scale them (to treat them as plain byte offsets
+// in memory, so that perhaps the loaded values are unaligned). The
+// scaled instructions' address operand in assembly looks like
+// [Rn,Qm,UXTW #2] or similar.
+
+// Base class.
+class MVE_VLDRSTR_rq<MVE_ldst_direction dir, MVE_memsz memsz, bit U,
+                     bits<2> size, bit os, string asm, string suffix, int shift>
+  : MVE_VLDRSTR_base<dir, U, 0b0, 0b0, 0, dir.Oops,
+                     !con(dir.Iops, (ins mve_addr_rq_shift<shift>:$addr)),
+                     asm, suffix, "$Qd, $addr", dir.cstr> {
+  bits<7> addr;
+  let Inst{23} = 0b1;
+  let Inst{19-16} = addr{6-3};
+  let Inst{8-7} = size;
+  let Inst{6} = memsz.encoding{1};
+  let Inst{5} = 0;
+  let Inst{4} = memsz.encoding{0};
+  let Inst{3-1} = addr{2-0};
+  let Inst{0} = os;
+}
+
+// Multiclass that defines the scaled and unscaled versions of an
+// instruction, when the memory size is wider than a byte. The scaled
+// version gets the default name like MVE_VLDRBU16_rq; the unscaled /
+// potentially unaligned version gets a "_u" suffix, e.g.
+// MVE_VLDRBU16_rq_u.
+multiclass MVE_VLDRSTR_rq_w<MVE_ldst_direction dir, MVE_memsz memsz,
+                            string asm, string suffix, bit U, bits<2> size> {
+  def _u : MVE_VLDRSTR_rq<dir, memsz, U, size, 0, asm, suffix, 0>;
+  def "" : MVE_VLDRSTR_rq<dir, memsz, U, size, 1, asm, suffix, memsz.shift>;
+}
+
+// Trivial multiclass (only defining one thing), with the same API as
+// the previous one, for use when the memory size is one byte, so
+// there's no 'scaled' version of the instruction at all. (This is
+// encoded as if it were unscaled, but named in the default way with
+// no _u suffix.)
+multiclass MVE_VLDRSTR_rq_b<MVE_ldst_direction dir, MVE_memsz memsz,
+                            string asm, string suffix, bit U, bits<2> size> {
+  def "" : MVE_VLDRSTR_rq<dir, memsz, U, size, 0, asm, suffix, 0>;
+}
+
+// Actually define all the loads and stores in this family.
+
+defm MVE_VLDRBU8_rq : MVE_VLDRSTR_rq_b<MVE_ld, MVE_memB, "vldrb","u8",  1,0b00>;
+defm MVE_VLDRBU16_rq: MVE_VLDRSTR_rq_b<MVE_ld, MVE_memB, "vldrb","u16", 1,0b01>;
+defm MVE_VLDRBS16_rq: MVE_VLDRSTR_rq_b<MVE_ld, MVE_memB, "vldrb","s16", 0,0b01>;
+defm MVE_VLDRBU32_rq: MVE_VLDRSTR_rq_b<MVE_ld, MVE_memB, "vldrb","u32", 1,0b10>;
+defm MVE_VLDRBS32_rq: MVE_VLDRSTR_rq_b<MVE_ld, MVE_memB, "vldrb","s32", 0,0b10>;
+
+defm MVE_VLDRHU16_rq: MVE_VLDRSTR_rq_w<MVE_ld, MVE_memH, "vldrh","u16", 1,0b01>;
+defm MVE_VLDRHU32_rq: MVE_VLDRSTR_rq_w<MVE_ld, MVE_memH, "vldrh","u32", 1,0b10>;
+defm MVE_VLDRHS32_rq: MVE_VLDRSTR_rq_w<MVE_ld, MVE_memH, "vldrh","s32", 0,0b10>;
+defm MVE_VLDRWU32_rq: MVE_VLDRSTR_rq_w<MVE_ld, MVE_memW, "vldrw","u32", 1,0b10>;
+defm MVE_VLDRDU64_rq: MVE_VLDRSTR_rq_w<MVE_ld, MVE_memD, "vldrd","u64", 1,0b11>;
+
+defm MVE_VSTRB8_rq  : MVE_VLDRSTR_rq_b<MVE_st, MVE_memB, "vstrb","8",   0,0b00>;
+defm MVE_VSTRB16_rq : MVE_VLDRSTR_rq_b<MVE_st, MVE_memB, "vstrb","16",  0,0b01>;
+defm MVE_VSTRB32_rq : MVE_VLDRSTR_rq_b<MVE_st, MVE_memB, "vstrb","32",  0,0b10>;
+
+defm MVE_VSTRH16_rq : MVE_VLDRSTR_rq_w<MVE_st, MVE_memH, "vstrh","16",  0,0b01>;
+defm MVE_VSTRH32_rq : MVE_VLDRSTR_rq_w<MVE_st, MVE_memH, "vstrh","32",  0,0b10>;
+defm MVE_VSTRW32_rq : MVE_VLDRSTR_rq_w<MVE_st, MVE_memW, "vstrw","32",  0,0b10>;
+defm MVE_VSTRD64_rq : MVE_VLDRSTR_rq_w<MVE_st, MVE_memD, "vstrd","64",  0,0b11>;
+
+// Gather loads / scatter stores whose address operand is of the form
+// [Qm,#imm], i.e. a vector containing a full base address for each
+// loaded item, plus an immediate offset applied consistently to all
+// of them. ('Load/store the same field from this vector of pointers
+// to a structure type.')
+//
+// This family requires the vector lane size to be at least 32 bits
+// (so there's room for an address in each lane at all). It has no
+// widening/narrowing variants. But it does support preindex
+// writeback, in which the address vector is updated to hold the
+// addresses actually loaded from.
+
+// Base class.
+class MVE_VLDRSTR_qi<MVE_ldst_direction dir, MVE_memsz memsz, bit W, dag wbops,
+                     string asm, string wbAsm, string suffix, string cstr = "">
+  : MVE_VLDRSTR_base<dir, 1, 1, W, 1, !con(wbops, dir.Oops),
+                     !con(dir.Iops, (ins mve_addr_q_shift<memsz.shift>:$addr)),
+                     asm, suffix, "$Qd, $addr" # wbAsm, cstr # dir.cstr> {
+  bits<11> addr;
+  let Inst{23} = addr{7};
+  let Inst{19-17} = addr{10-8};
+  let Inst{16} = 0;
+  let Inst{8} = memsz.encoding{0}; // enough to distinguish 32- from 64-bit
+  let Inst{7} = 0;
+  let Inst{6-0} = addr{6-0};
+}
+
+// Multiclass that generates the non-writeback and writeback variants.
+multiclass MVE_VLDRSTR_qi_m<MVE_ldst_direction dir, MVE_memsz memsz,
+                            string asm, string suffix> {
+  def ""   : MVE_VLDRSTR_qi<dir, memsz, 0, (outs),          asm, "",  suffix>;
+  def _pre : MVE_VLDRSTR_qi<dir, memsz, 1, (outs MQPR:$wb), asm, "!", suffix,
+                            "$addr.base = $wb"> {
+    let DecoderMethod="DecodeMVE_MEM_3_pre<"#memsz.shift#">";
+  }
+}
+
+// Actual instruction definitions.
+defm MVE_VLDRWU32_qi: MVE_VLDRSTR_qi_m<MVE_ld, MVE_memW, "vldrw", "u32">;
+defm MVE_VLDRDU64_qi: MVE_VLDRSTR_qi_m<MVE_ld, MVE_memD, "vldrd", "u64">;
+defm MVE_VSTRW32_qi:  MVE_VLDRSTR_qi_m<MVE_st, MVE_memW, "vstrw", "32">;
+defm MVE_VSTRD64_qi:  MVE_VLDRSTR_qi_m<MVE_st, MVE_memD, "vstrd", "64">;
+
+// Define aliases for all the instructions where memory size and
+// vector lane size are the same. These are mnemonic aliases, so they
+// apply consistently across all of the above families - contiguous
+// loads, and both the rq and qi types of gather/scatter.
+//
+// Rationale: As long as you're loading (for example) 16-bit memory
+// values into 16-bit vector lanes, you can think of them as signed or
+// unsigned integers, fp16 or just raw 16-bit blobs and it makes no
+// difference. So we permit all of vldrh.16, vldrh.u16, vldrh.s16,
+// vldrh.f16 and treat them all as equivalent to the canonical
+// spelling (which happens to be .u16 for loads, and just .16 for
+// stores).
+
+foreach vpt_cond = ["", "t", "e"] in
+foreach memsz = [MVE_memB, MVE_memH, MVE_memW, MVE_memD] in
+foreach suffix = memsz.suffixes in {
+
+  // These foreaches are conceptually ifs, implemented by iterating a
+  // dummy variable over a list with 0 or 1 elements depending on the
+  // condition. The idea is to iterate over _nearly_ all the suffixes
+  // in memsz.suffixes, but omit the one we want all the others to alias.
+
+  foreach _ = !if(!ne(suffix, memsz.CanonLoadSuffix), [1], []<int>) in
+  def : MnemonicAlias<
+    "vldr" # memsz.MnemonicLetter # vpt_cond # suffix,
+    "vldr" # memsz.MnemonicLetter # vpt_cond # memsz.CanonLoadSuffix>;
+
+  foreach _ = !if(!ne(suffix, memsz.CanonStoreSuffix), [1], []<int>) in
+  def : MnemonicAlias<
+    "vstr" # memsz.MnemonicLetter # vpt_cond # suffix,
+    "vstr" # memsz.MnemonicLetter # vpt_cond # memsz.CanonStoreSuffix>;
+}
+
+// end of MVE predicable load/store
+
 class MVE_VPT<string suffix, bits<2> size, dag iops, string asm, list<dag> pattern=[]>
   : MVE_MI<(outs ), iops, NoItinerary, !strconcat("vpt", "${Mk}", ".", suffix), asm, "", pattern> {
   bits<3> fc;
   bits<4> Mk;
   bits<3> Qn;
 
   let Inst{31-23} = 0b111111100;
   let Inst{22} = Mk{3};
def MVE_VPST : MVE_MI<(outs ), (ins vpt_mask:$Mk), NoItinerary,
   let Inst{21-16} = 0b110001;
   let Inst{15-13} = Mk{2-0};
   let Inst{12-0} = 0b0111101001101;
   let Unpredictable{12} = 0b1;
   let Unpredictable{7} = 0b1;
   let Unpredictable{5} = 0b1;
 
   let Defs = [P0];
 }

llvm/lib/Target/ARM/ARMInstrThumb2.td

 
 // ADR instruction labels.
 def t2adrlabel : Operand<i32> {
   let EncoderMethod = "getT2AdrLabelOpValue";
   let PrintMethod = "printAdrLabelOperand<0>";
 }
 
 // t2addrmode_posimm8  := reg + imm8
-def MemPosImm8OffsetAsmOperand : AsmOperandClass {let Name="MemPosImm8Offset";}
+def MemPosImm8OffsetAsmOperand : AsmOperandClass {
+  let Name="MemPosImm8Offset";
+  let RenderMethod = "addMemImmOffsetOperands";
+}
 def t2addrmode_posimm8 : MemOperand {
   let PrintMethod = "printT2AddrModeImm8Operand<false>";
-  let EncoderMethod = "getT2AddrModeImm8OpValue";
+  let EncoderMethod = "getT2AddrModeImmOpValue<8,0>";
   let DecoderMethod = "DecodeT2AddrModeImm8";
   let ParserMatchClass = MemPosImm8OffsetAsmOperand;
   let MIOperandInfo = (ops GPR:$base, i32imm:$offsimm);
 }
 
 // t2addrmode_negimm8  := reg - imm8
-def MemNegImm8OffsetAsmOperand : AsmOperandClass {let Name="MemNegImm8Offset";}
+def MemNegImm8OffsetAsmOperand : AsmOperandClass {
+  let Name="MemNegImm8Offset";
+  let RenderMethod = "addMemImmOffsetOperands";
+}
 def t2addrmode_negimm8 : MemOperand,
                       ComplexPattern<i32, 2, "SelectT2AddrModeImm8", []> {
   let PrintMethod = "printT2AddrModeImm8Operand<false>";
-  let EncoderMethod = "getT2AddrModeImm8OpValue";
+  let EncoderMethod = "getT2AddrModeImmOpValue<8,0>";
   let DecoderMethod = "DecodeT2AddrModeImm8";
   let ParserMatchClass = MemNegImm8OffsetAsmOperand;
   let MIOperandInfo = (ops GPR:$base, i32imm:$offsimm);
 }
 
 // t2addrmode_imm8  := reg +/- imm8
-def MemImm8OffsetAsmOperand : AsmOperandClass { let Name = "MemImm8Offset"; }
+def MemImm8OffsetAsmOperand : AsmOperandClass {
+  let Name = "MemImm8Offset";
+  let RenderMethod = "addMemImmOffsetOperands";
+}
 class T2AddrMode_Imm8 : MemOperand,
                         ComplexPattern<i32, 2, "SelectT2AddrModeImm8", []> {
-  let EncoderMethod = "getT2AddrModeImm8OpValue";
+  let EncoderMethod = "getT2AddrModeImmOpValue<8,0>";
   let DecoderMethod = "DecodeT2AddrModeImm8";
   let ParserMatchClass = MemImm8OffsetAsmOperand;
   let MIOperandInfo = (ops GPR:$base, i32imm:$offsimm);
 }
 
 def t2addrmode_imm8 : T2AddrMode_Imm8 {
   let PrintMethod = "printT2AddrModeImm8Operand<false>";
 }
def t2am_imm7s4_offset : MemOperand {
   // They are printed the same way as the imm8 version
   let PrintMethod = "printT2AddrModeImm8s4OffsetOperand";
   let ParserMatchClass = t2am_imm7s4_offset_asmoperand;
   let EncoderMethod = "getT2ScaledImmOpValue<7,2>";
   let DecoderMethod = "DecodeT2Imm7S4";
 }
 
 // t2addrmode_imm0_1020s4  := reg + (imm8 << 2)
 def MemImm0_1020s4OffsetAsmOperand : AsmOperandClass {
   let Name = "MemImm0_1020s4Offset";
 }
 def t2addrmode_imm0_1020s4 : MemOperand,
                          ComplexPattern<i32, 2, "SelectT2AddrModeExclusive"> {
   let PrintMethod = "printT2AddrModeImm0_1020s4Operand";
   let EncoderMethod = "getT2AddrModeImm0_1020s4OpValue";
   let DecoderMethod = "DecodeT2AddrModeImm0_1020s4";
   let ParserMatchClass = MemImm0_1020s4OffsetAsmOperand;

llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp

public:
   bool isThumbMemPC() const {
     int64_t Val = 0;
     if (isImm()) {
       if (isa<MCSymbolRefExpr>(Imm.Val)) return true;
       const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Imm.Val);
       if (!CE) return false;
       Val = CE->getValue();
     }
-    else if (isMem()) {
+    else if (isGPRMem()) {
       if(!Memory.OffsetImm || Memory.OffsetRegNum) return false;
       if(Memory.BaseRegNum != ARM::PC) return false;
       Val = Memory.OffsetImm->getValue();
     }
     else return false;
     return ((Val % 4) == 0) && (Val >= 0) && (Val <= 1020);
   }
 
public:
   template<int64_t N, int64_t M>
   bool isImmediateS4() const {
     if (!isImm()) return false;
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
     if (!CE) return false;
     int64_t Value = CE->getValue();
     return ((Value & 3) == 0) && Value >= N && Value <= M;
   }
-
+  template<int64_t N, int64_t M>
+  bool isImmediateS2() const {
+    if (!isImm()) return false;
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    if (!CE) return false;
+    int64_t Value = CE->getValue();
+    return ((Value & 1) == 0) && Value >= N && Value <= M;
+  }
   bool isFBits16() const {
     return isImmediate<0, 17>();
   }
   bool isFBits32() const {
     return isImmediate<1, 33>();
   }
   bool isImm8s4() const {
     return isImmediateS4<-1020, 1020>();
   }
   bool isImm7s4() const {
     return isImmediateS4<-508, 508>();
   }
+  bool isImm7Shift0() const {
+    return isImmediate<-127, 127>();
+  }
+  bool isImm7Shift1() const {
+    return isImmediateS2<-255, 255>();
+  }
+  bool isImm7Shift2() const {
+    return isImmediateS4<-511, 511>();
+  }
+  bool isImm7() const {
+    return isImmediate<-127, 127>();
+  }
   bool isImm0_1020s4() const {
     return isImmediateS4<0, 1020>();
   }
   bool isImm0_508s4() const {
     return isImmediateS4<0, 508>();
   }
   bool isImm0_508s4Neg() const {
     if (!isImm()) return false;
public:
   bool isSPRRegList() const { return Kind == k_SPRRegisterList; }
   bool isFPSRegListWithVPR() const { return Kind == k_FPSRegisterListWithVPR; }
   bool isFPDRegListWithVPR() const { return Kind == k_FPDRegisterListWithVPR; }
   bool isToken() const override { return Kind == k_Token; }
   bool isMemBarrierOpt() const { return Kind == k_MemBarrierOpt; }
   bool isInstSyncBarrierOpt() const { return Kind == k_InstSyncBarrierOpt; }
   bool isTraceSyncBarrierOpt() const { return Kind == k_TraceSyncBarrierOpt; }
   bool isMem() const override {
+      return isGPRMem() || isMVEMem();
+  }
+  bool isMVEMem() const {
+    if (Kind != k_Memory)
+      return false;
+    if (Memory.BaseRegNum &&
+        !ARMMCRegisterClasses[ARM::GPRRegClassID].contains(Memory.BaseRegNum) &&
+        !ARMMCRegisterClasses[ARM::MQPRRegClassID].contains(Memory.BaseRegNum))
+      return false;
+    if (Memory.OffsetRegNum &&
+        !ARMMCRegisterClasses[ARM::MQPRRegClassID].contains(
+            Memory.OffsetRegNum))
+      return false;
+    return true;
+  }
+  bool isGPRMem() const {
     if (Kind != k_Memory)
       return false;
     if (Memory.BaseRegNum &&
         !ARMMCRegisterClasses[ARM::GPRRegClassID].contains(Memory.BaseRegNum))
       return false;
     if (Memory.OffsetRegNum &&
         !ARMMCRegisterClasses[ARM::GPRRegClassID].contains(Memory.OffsetRegNum))
       return false;
public:
   bool isPostIdxRegShifted() const {
     return Kind == k_PostIndexRegister &&
            ARMMCRegisterClasses[ARM::GPRRegClassID].contains(PostIdxReg.RegNum);
   }
   bool isPostIdxReg() const {
     return isPostIdxRegShifted() && PostIdxReg.ShiftTy == ARM_AM::no_shift;
   }
   bool isMemNoOffset(bool alignOK = false, unsigned Alignment = 0) const {
-    if (!isMem())
+    if (!isGPRMem())
       return false;
     // No offset of any kind.
     return Memory.OffsetRegNum == 0 && Memory.OffsetImm == nullptr &&
      (alignOK || Memory.Alignment == Alignment);
   }
   bool isMemNoOffsetT2(bool alignOK = false, unsigned Alignment = 0) const {
-    if (!isMem())
+    if (!isGPRMem())
       return false;
 
     if (!ARMMCRegisterClasses[ARM::GPRnopcRegClassID].contains(
             Memory.BaseRegNum))
       return false;
 
     // No offset of any kind.
     return Memory.OffsetRegNum == 0 && Memory.OffsetImm == nullptr &&
      (alignOK || Memory.Alignment == Alignment);
   }
   bool isMemNoOffsetT2NoSp(bool alignOK = false, unsigned Alignment = 0) const {
-    if (!isMem())
+    if (!isGPRMem())
       return false;
 
     if (!ARMMCRegisterClasses[ARM::rGPRRegClassID].contains(
             Memory.BaseRegNum))
       return false;
 
     // No offset of any kind.
     return Memory.OffsetRegNum == 0 && Memory.OffsetImm == nullptr &&
      (alignOK || Memory.Alignment == Alignment);
   }
+  bool isMemNoOffsetT(bool alignOK = false, unsigned Alignment = 0) const {
+    if (!isGPRMem())
+      return false;
+
+    if (!ARMMCRegisterClasses[ARM::tGPRRegClassID].contains(
+            Memory.BaseRegNum))
+      return false;
+
+    // No offset of any kind.
+    return Memory.OffsetRegNum == 0 && Memory.OffsetImm == nullptr &&
+     (alignOK || Memory.Alignment == Alignment);
+  }
   bool isMemPCRelImm12() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Base register must be PC.
     if (Memory.BaseRegNum != ARM::PC)
       return false;
     // Immediate offset in range [-4095, 4095].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return (Val > -4096 && Val < 4096) ||
bool isAlignedMemory64or128or256() const {
     if (isMemNoOffset(false, 16)) // alignment in bytes for 128-bits is 16.
       return true;
     if (isMemNoOffset(false, 32)) // alignment in bytes for 256-bits is 32.
       return true;
     return isMemNoOffset(false, 0);
   }
 
   bool isAddrMode2() const {
-    if (!isMem() || Memory.Alignment != 0) return false;
+    if (!isGPRMem() || Memory.Alignment != 0) return false;
     // Check for register offset.
     if (Memory.OffsetRegNum) return true;
     // Immediate offset in range [-4095, 4095].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return Val > -4096 && Val < 4096;
   }
 
   bool isAM2OffsetImm() const {
     if (!isImm()) return false;
     // Immediate offset in range [-4095, 4095].
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
     if (!CE) return false;
     int64_t Val = CE->getValue();
     return (Val == std::numeric_limits<int32_t>::min()) ||
            (Val > -4096 && Val < 4096);
   }
 
   bool isAddrMode3() const {
     // If we have an immediate that's not a constant, treat it as a label
     // reference needing a fixup. If it is a constant, it's something else
     // and we reject it.
     if (isImm() && !isa<MCConstantExpr>(getImm()))
       return true;
-    if (!isMem() || Memory.Alignment != 0) return false;
+    if (!isGPRMem() || Memory.Alignment != 0) return false;
     // No shifts are legal for AM3.
     if (Memory.ShiftType != ARM_AM::no_shift) return false;
     // Check for register offset.
     if (Memory.OffsetRegNum) return true;
     // Immediate offset in range [-255, 255].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     // The #-0 offset is encoded as std::numeric_limits<int32_t>::min(), and we
public:
   }
 
   bool isAddrMode5() const {
     // If we have an immediate that's not a constant, treat it as a label
     // reference needing a fixup. If it is a constant, it's something else
     // and we reject it.
     if (isImm() && !isa<MCConstantExpr>(getImm()))
       return true;
-    if (!isMem() || Memory.Alignment != 0) return false;
+    if (!isGPRMem() || Memory.Alignment != 0) return false;
     // Check for register offset.
     if (Memory.OffsetRegNum) return false;
     // Immediate offset in range [-1020, 1020] and a multiple of 4.
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return (Val >= -1020 && Val <= 1020 && ((Val & 3) == 0)) ||
       Val == std::numeric_limits<int32_t>::min();
   }
 
   bool isAddrMode5FP16() const {
     // If we have an immediate that's not a constant, treat it as a label
     // reference needing a fixup. If it is a constant, it's something else
     // and we reject it.
     if (isImm() && !isa<MCConstantExpr>(getImm()))
       return true;
-    if (!isMem() || Memory.Alignment != 0) return false;
+    if (!isGPRMem() || Memory.Alignment != 0) return false;
     // Check for register offset.
     if (Memory.OffsetRegNum) return false;
     // Immediate offset in range [-510, 510] and a multiple of 2.
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return (Val >= -510 && Val <= 510 && ((Val & 1) == 0)) ||
            Val == std::numeric_limits<int32_t>::min();
   }
 
   bool isMemTBB() const {
-    if (!isMem() || !Memory.OffsetRegNum || Memory.isNegative ||
+    if (!isGPRMem() || !Memory.OffsetRegNum || Memory.isNegative ||
         Memory.ShiftType != ARM_AM::no_shift || Memory.Alignment != 0)
       return false;
     return true;
   }
 
   bool isMemTBH() const {
-    if (!isMem() || !Memory.OffsetRegNum || Memory.isNegative ||
+    if (!isGPRMem() || !Memory.OffsetRegNum || Memory.isNegative ||
         Memory.ShiftType != ARM_AM::lsl || Memory.ShiftImm != 1 ||
         Memory.Alignment != 0 )
       return false;
     return true;
   }
 
   bool isMemRegOffset() const {
-    if (!isMem() || !Memory.OffsetRegNum || Memory.Alignment != 0)
+    if (!isGPRMem() || !Memory.OffsetRegNum || Memory.Alignment != 0)
       return false;
     return true;
   }
 
   bool isT2MemRegOffset() const {
-    if (!isMem() || !Memory.OffsetRegNum || Memory.isNegative ||
+    if (!isGPRMem() || !Memory.OffsetRegNum || Memory.isNegative ||
         Memory.Alignment != 0 || Memory.BaseRegNum == ARM::PC)
       return false;
     // Only lsl #{0, 1, 2, 3} allowed.
     if (Memory.ShiftType == ARM_AM::no_shift)
       return true;
     if (Memory.ShiftType != ARM_AM::lsl || Memory.ShiftImm > 3)
       return false;
     return true;
   }
 
   bool isMemThumbRR() const {
     // Thumb reg+reg addressing is simple. Just two registers, a base and
     // an offset. No shifts, negations or any other complicating factors.
-    if (!isMem() || !Memory.OffsetRegNum || Memory.isNegative ||
+    if (!isGPRMem() || !Memory.OffsetRegNum || Memory.isNegative ||
         Memory.ShiftType != ARM_AM::no_shift || Memory.Alignment != 0)
       return false;
     return isARMLowRegister(Memory.BaseRegNum) &&
       (!Memory.OffsetRegNum || isARMLowRegister(Memory.OffsetRegNum));
   }
 
   bool isMemThumbRIs4() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 ||
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 ||
         !isARMLowRegister(Memory.BaseRegNum) || Memory.Alignment != 0)
       return false;
     // Immediate offset, multiple of 4 in range [0, 124].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return Val >= 0 && Val <= 124 && (Val % 4) == 0;
   }
 
   bool isMemThumbRIs2() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 ||
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 ||
         !isARMLowRegister(Memory.BaseRegNum) || Memory.Alignment != 0)
       return false;
     // Immediate offset, multiple of 4 in range [0, 62].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return Val >= 0 && Val <= 62 && (Val % 2) == 0;
   }
 
   bool isMemThumbRIs1() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 ||
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 ||
         !isARMLowRegister(Memory.BaseRegNum) || Memory.Alignment != 0)
       return false;
     // Immediate offset in range [0, 31].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return Val >= 0 && Val <= 31;
   }
 
   bool isMemThumbSPI() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 ||
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 ||
         Memory.BaseRegNum != ARM::SP || Memory.Alignment != 0)
       return false;
     // Immediate offset, multiple of 4 in range [0, 1020].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return Val >= 0 && Val <= 1020 && (Val % 4) == 0;
   }
 
   bool isMemImm8s4Offset() const {
     // If we have an immediate that's not a constant, treat it as a label
     // reference needing a fixup. If it is a constant, it's something else
     // and we reject it.
     if (isImm() && !isa<MCConstantExpr>(getImm()))
       return true;
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Immediate offset a multiple of 4 in range [-1020, 1020].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     // Special case, #-0 is std::numeric_limits<int32_t>::min().
     return (Val >= -1020 && Val <= 1020 && (Val & 3) == 0) ||
            Val == std::numeric_limits<int32_t>::min();
   }
   bool isMemImm7s4Offset() const {
     // If we have an immediate that's not a constant, treat it as a label
     // reference needing a fixup. If it is a constant, it's something else
     // and we reject it.
     if (isImm() && !isa<MCConstantExpr>(getImm()))
       return true;
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0 ||
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0 ||
         !ARMMCRegisterClasses[ARM::GPRnopcRegClassID].contains(
             Memory.BaseRegNum))
       return false;
     // Immediate offset a multiple of 4 in range [-508, 508].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     // Special case, #-0 is INT32_MIN.
     return (Val >= -508 && Val <= 508 && (Val & 3) == 0) || Val == INT32_MIN;
   }
   bool isMemImm0_1020s4Offset() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Immediate offset a multiple of 4 in range [0, 1020].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return Val >= 0 && Val <= 1020 && (Val & 3) == 0;
   }
 
   bool isMemImm8Offset() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Base reg of PC isn't allowed for these encodings.
     if (Memory.BaseRegNum == ARM::PC) return false;
     // Immediate offset in range [-255, 255].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return (Val == std::numeric_limits<int32_t>::min()) ||
            (Val > -256 && Val < 256);
   }
 
+  template<unsigned Bits, unsigned RegClassID>
+  bool isMemImm7ShiftedOffset() const {
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0 ||
+        !ARMMCRegisterClasses[RegClassID].contains(Memory.BaseRegNum))
+      return false;
+
+    // Expect an immediate offset equal to an element of the range
+    // [-127, 127], shifted left by Bits.
+
+    if (!Memory.OffsetImm) return true;
+    int64_t Val = Memory.OffsetImm->getValue();
+
+    // INT32_MIN is a special-case value (indicating the encoding with
+    // zero offset and the subtract bit set)
+    if (Val == INT32_MIN)
+      return true;
+
+    unsigned Divisor = 1U << Bits;
+
+    // Check that the low bits are zero
+    if (Val % Divisor != 0)
+      return false;
+
+    // Check that the remaining offset is within range.
+    Val /= Divisor;
+    return (Val >= -127 && Val <= 127);
+  }
+
+  template <int shift> bool isMemRegRQOffset() const {
+    if (!isMVEMem() || Memory.OffsetImm != 0 || Memory.Alignment != 0)
+      return false;
+
+    if (!ARMMCRegisterClasses[ARM::GPRnopcRegClassID].contains(
+            Memory.BaseRegNum))
+      return false;
+    if (!ARMMCRegisterClasses[ARM::MQPRRegClassID].contains(
+            Memory.OffsetRegNum))
+      return false;
+
+    if (shift == 0 && Memory.ShiftType != ARM_AM::no_shift)
+      return false;
+
+    if (shift > 0 &&
+        (Memory.ShiftType != ARM_AM::uxtw || Memory.ShiftImm != shift))
+      return false;
+
+    return true;
+  }
+
+  template <int shift> bool isMemRegQOffset() const {
+    if (!isMVEMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+      return false;
+
+    if (!ARMMCRegisterClasses[ARM::MQPRRegClassID].contains(
+            Memory.BaseRegNum))
+      return false;
+
+    if(!Memory.OffsetImm) return true;
+    static_assert(shift < 56,
+                  "Such that we dont shift by a value higher than 62");
+    int64_t Val = Memory.OffsetImm->getValue();
+
+    // The value must be a multiple of (1 << shift)
+    if ((Val & ((1U << shift) - 1)) != 0)
+      return false;
+
+    // And be in the right range, depending on the amount that it is shifted
+    // by.  Shift 0, is equal to 7 unsigned bits, the sign bit is set
+    // separately.
+    int64_t Range = (1U << (7+shift)) - 1;
+    return (Val == INT32_MIN) || (Val > -Range && Val < Range);
+  }
+
   bool isMemPosImm8Offset() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Immediate offset in range [0, 255].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return Val >= 0 && Val < 256;
   }
 
   bool isMemNegImm8Offset() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Base reg of PC isn't allowed for these encodings.
     if (Memory.BaseRegNum == ARM::PC) return false;
     // Immediate offset in range [-255, -1].
     if (!Memory.OffsetImm) return false;
     int64_t Val = Memory.OffsetImm->getValue();
     return (Val == std::numeric_limits<int32_t>::min()) ||
            (Val > -256 && Val < 0);
   }
 
   bool isMemUImm12Offset() const {
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Immediate offset in range [0, 4095].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return (Val >= 0 && Val < 4096);
   }
 
   bool isMemImm12Offset() const {
     // If we have an immediate that's not a constant, treat it as a label
     // reference needing a fixup. If it is a constant, it's something else
     // and we reject it.
 
     if (isImm() && !isa<MCConstantExpr>(getImm()))
       return true;
 
-    if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
+    if (!isGPRMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
       return false;
     // Immediate offset in range [-4095, 4095].
     if (!Memory.OffsetImm) return true;
     int64_t Val = Memory.OffsetImm->getValue();
     return (Val > -4096 && Val < 4096) ||
            (Val == std::numeric_limits<int32_t>::min());
   }
 

   void addImm7s4Operands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");
     // FIXME: We really want to scale the value here, but the VSTR/VLDR_VSYSR
     // instruction don't encode operands that way yet.
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
     Inst.addOperand(MCOperand::createImm(CE->getValue()));
   }
 
+  void addImm7Shift0Operands(MCInst &Inst, unsigned N) const {
+    assert(N == 1 && "Invalid number of operands!");
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    assert(CE != nullptr && "Invalid operand type!");
+    Inst.addOperand(MCOperand::createImm(CE->getValue()));
+  }
+
+  void addImm7Shift1Operands(MCInst &Inst, unsigned N) const {
+    assert(N == 1 && "Invalid number of operands!");
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    assert(CE != nullptr && "Invalid operand type!");
+    Inst.addOperand(MCOperand::createImm(CE->getValue()));
+  }
+
+  void addImm7Shift2Operands(MCInst &Inst, unsigned N) const {
+    assert(N == 1 && "Invalid number of operands!");
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    assert(CE != nullptr && "Invalid operand type!");
+    Inst.addOperand(MCOperand::createImm(CE->getValue()));
+  }
+
+  void addImm7Operands(MCInst &Inst, unsigned N) const {
+    assert(N == 1 && "Invalid number of operands!");
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    assert(CE != nullptr && "Invalid operand type!");
+    Inst.addOperand(MCOperand::createImm(CE->getValue()));
+  }
+
   void addImm0_1020s4Operands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");
     // The immediate is scaled by four in the encoding and is stored
     // in the MCInst as such. Lop off the low two bits here.
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
     Inst.addOperand(MCOperand::createImm(CE->getValue() / 4));
   }
 
if (isImm()) {
 
       const MCSymbolRefExpr *SR = dyn_cast<MCSymbolRefExpr>(Imm.Val);
 
       assert(SR && "Unknown value type!");
       Inst.addOperand(MCOperand::createExpr(SR));
       return;
     }
 
-    assert(isMem()  && "Unknown value type!");
+    assert(isGPRMem()  && "Unknown value type!");
     assert(isa<MCConstantExpr>(Memory.OffsetImm) && "Unknown value type!");
     Inst.addOperand(MCOperand::createImm(Memory.OffsetImm->getValue()));
   }
 
   void addMemBarrierOptOperands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");
     Inst.addOperand(MCOperand::createImm(unsigned(getMemBarrierOpt())));
   }
void addMemNoOffsetT2Operands(MCInst &Inst, unsigned N) const {
     Inst.addOperand(MCOperand::createReg(Memory.BaseRegNum));
   }
 
   void addMemNoOffsetT2NoSpOperands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");
     Inst.addOperand(MCOperand::createReg(Memory.BaseRegNum));
   }
 
+  void addMemNoOffsetTOperands(MCInst &Inst, unsigned N) const {
+    assert(N == 1 && "Invalid number of operands!");
+    Inst.addOperand(MCOperand::createReg(Memory.BaseRegNum));
+  }
+
   void addMemPCRelImm12Operands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");
     int32_t Imm = Memory.OffsetImm->getValue();
     Inst.addOperand(MCOperand::createImm(Imm));
   }
 
   void addAdrLabelOperands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");

   void addMemImm0_1020s4OffsetOperands(MCInst &Inst, unsigned N) const {
     assert(N == 2 && "Invalid number of operands!");
     // The lower two bits are always zero and as such are not encoded.
     int32_t Val = Memory.OffsetImm ? Memory.OffsetImm->getValue() / 4 : 0;
     Inst.addOperand(MCOperand::createReg(Memory.BaseRegNum));
     Inst.addOperand(MCOperand::createImm(Val));
   }
 
-  void addMemImm8OffsetOperands(MCInst &Inst, unsigned N) const {
+  void addMemImmOffsetOperands(MCInst &Inst, unsigned N) const {
     assert(N == 2 && "Invalid number of operands!");
     int64_t Val = Memory.OffsetImm ? Memory.OffsetImm->getValue() : 0;
     Inst.addOperand(MCOperand::createReg(Memory.BaseRegNum));
     Inst.addOperand(MCOperand::createImm(Val));
   }
 
-  void addMemPosImm8OffsetOperands(MCInst &Inst, unsigned N) const {
-    addMemImm8OffsetOperands(Inst, N);
-  }
-
-  void addMemNegImm8OffsetOperands(MCInst &Inst, unsigned N) const {
-    addMemImm8OffsetOperands(Inst, N);
+  void addMemRegRQOffsetOperands(MCInst &Inst, unsigned N) const {
+    assert(N == 2 && "Invalid number of operands!");
+    Inst.addOperand(MCOperand::createReg(Memory.BaseRegNum));
+    Inst.addOperand(MCOperand::createReg(Memory.OffsetRegNum));
   }
 
   void addMemUImm12OffsetOperands(MCInst &Inst, unsigned N) const {
     assert(N == 2 && "Invalid number of operands!");
     // If this is an immediate, it's a label reference.
     if (isImm()) {
       addExpr(Inst, getImm());
       Inst.addOperand(MCOperand::createImm(0));
       return;
     }
 
ARMAsmParser::parseVectorList(OperandVector &Operands) {
   case AllLanes: {
     // Two-register operands have been converted to the
     // composite register classes.
     if (Count == 2 && !hasMVE()) {
       const MCRegisterClass *RC = (Spacing == 1) ?
         &ARMMCRegisterClasses[ARM::DPairRegClassID] :
         &ARMMCRegisterClasses[ARM::DPairSpcRegClassID];
       FirstReg = MRI->getMatchingSuperReg(FirstReg, ARM::dsub_0, RC);
     }
     auto Create = (LaneKind == NoLanes ? ARMOperand::CreateVectorList :
                    ARMOperand::CreateVectorListAllLanes);
     Operands.push_back(Create(FirstReg, Count, (Spacing == 2), S, E));
     break;
   }
   case IndexedLane:
     Operands.push_back(ARMOperand::CreateVectorListIndexed(FirstReg, Count,
                                                            LaneIndex,
                                                            (Spacing == 2),
bool ARMAsmParser::parseMemRegOffsetShift(ARM_AM::ShiftOpc &St,
   else if (ShiftName == "lsr" || ShiftName == "LSR")
     St = ARM_AM::lsr;
   else if (ShiftName == "asr" || ShiftName == "ASR")
     St = ARM_AM::asr;
   else if (ShiftName == "ror" || ShiftName == "ROR")
     St = ARM_AM::ror;
   else if (ShiftName == "rrx" || ShiftName == "RRX")
     St = ARM_AM::rrx;
+  else if (ShiftName == "uxtw" || ShiftName == "UXTW")
+    St= ARM_AM::uxtw;
   else
     return Error(Loc, "illegal shift operator");
   Parser.Lex(); // Eat shift type token.
 
   // rrx stands alone.
   Amount = 0;
   if (St != ARM_AM::rrx) {
     Loc = Parser.getTok().getLoc();
void ARMAsmParser::fixupGNULDRDAlias(StringRef Mnemonic,
   if (Operands.size() < 4)
     return;
 
   ARMOperand &Op2 = static_cast<ARMOperand &>(*Operands[2]);
   ARMOperand &Op3 = static_cast<ARMOperand &>(*Operands[3]);
 
   if (!Op2.isReg())
     return;
-  if (!Op3.isMem())
+  if (!Op3.isGPRMem())
     return;
 
   const MCRegisterClass &GPR = MRI->getRegClass(ARM::GPRRegClassID);
   if (!GPR.contains(Op2.getReg()))
     return;
 
   unsigned RtEncoding = MRI->getEncodingValue(Op2.getReg());
   if (!isThumb() && (RtEncoding & 1)) {
case ARM::t2LDRSB_POST: {
     const unsigned Rt = MRI->getEncodingValue(Inst.getOperand(0).getReg());
     const unsigned Rn = MRI->getEncodingValue(Inst.getOperand(2).getReg());
 
     if (Rt == Rn)
       return Error(Operands[3]->getStartLoc(),
                    "destination register and base register can't be identical");
     return false;
   }
+
+  case ARM::MVE_VLDRBU8_rq:
+  case ARM::MVE_VLDRBU16_rq:
+  case ARM::MVE_VLDRBS16_rq:
+  case ARM::MVE_VLDRBU32_rq:
+  case ARM::MVE_VLDRBS32_rq:
+  case ARM::MVE_VLDRHU16_rq:
+  case ARM::MVE_VLDRHU16_rq_u:
+  case ARM::MVE_VLDRHU32_rq:
+  case ARM::MVE_VLDRHU32_rq_u:
+  case ARM::MVE_VLDRHS32_rq:
+  case ARM::MVE_VLDRHS32_rq_u:
+  case ARM::MVE_VLDRWU32_rq:
+  case ARM::MVE_VLDRWU32_rq_u:
+  case ARM::MVE_VLDRDU64_rq:
+  case ARM::MVE_VLDRDU64_rq_u:
+  case ARM::MVE_VLDRWU32_qi:
+  case ARM::MVE_VLDRWU32_qi_pre:
+  case ARM::MVE_VLDRDU64_qi:
+  case ARM::MVE_VLDRDU64_qi_pre: {
+    // Qd must be different from Qm.
+    unsigned QdIdx = 0, QmIdx = 2;
+    bool QmIsPointer = false;
+    switch (Opcode) {
+    case ARM::MVE_VLDRWU32_qi:
+    case ARM::MVE_VLDRDU64_qi:
+      QmIdx = 1;
+      QmIsPointer = true;
+      break;
+    case ARM::MVE_VLDRWU32_qi_pre:
+    case ARM::MVE_VLDRDU64_qi_pre:
+      QdIdx = 1;
+      QmIsPointer = true;
+      break;
+    }
+
+    const unsigned Qd = MRI->getEncodingValue(Inst.getOperand(QdIdx).getReg());
+    const unsigned Qm = MRI->getEncodingValue(Inst.getOperand(QmIdx).getReg());
+
+    if (Qd == Qm) {
+      return Error(Operands[3]->getStartLoc(),
+                   Twine("destination vector register and vector ") +
+                   (QmIsPointer ? "pointer" : "offset") +
+                   " register can't be identical");
+    }
+    return false;
+  }
+
   case ARM::SBFX:
   case ARM::t2SBFX:
   case ARM::UBFX:
   case ARM::t2UBFX: {
     // Width must be in range [1, 32-lsb].
     unsigned LSB = Inst.getOperand(2).getImm();
     unsigned Widthm1 = Inst.getOperand(3).getImm();
     if (Widthm1 >= 32 - LSB)

llvm/lib/Target/ARM/Disassembler/ARMDisassembler.cpp

 static DecodeStatus DecodeShiftRight32Imm(MCInst &Inst, unsigned Val,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeShiftRight64Imm(MCInst &Inst, unsigned Val,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeTBLInstruction(MCInst &Inst, unsigned Insn,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodePostIdxReg(MCInst &Inst, unsigned Insn,
                                uint64_t Address, const void *Decoder);
+static DecodeStatus DecodeMveAddrModeRQ(MCInst &Inst, unsigned Insn,
+                               uint64_t Address, const void *Decoder);
+template<int shift>
+static DecodeStatus DecodeMveAddrModeQ(MCInst &Inst, unsigned Insn,
+                               uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeCoprocessor(MCInst &Inst, unsigned Insn,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeMemBarrierOption(MCInst &Inst, unsigned Insn,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeInstSyncBarrierOption(MCInst &Inst, unsigned Insn,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeMSRMask(MCInst &Inst, unsigned Insn,
                                uint64_t Address, const void *Decoder);
static DecodeStatus DecodeT2AddrModeImm8s4(MCInst &Inst, unsigned Val,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeT2AddrModeImm7s4(MCInst &Inst, unsigned Val,
                                            uint64_t Address,
                                            const void *Decoder);
 static DecodeStatus DecodeT2AddrModeImm0_1020s4(MCInst &Inst,unsigned Val,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeT2Imm8(MCInst &Inst, unsigned Val,
                                uint64_t Address, const void *Decoder);
+template<int shift>
+static DecodeStatus DecodeT2Imm7(MCInst &Inst, unsigned Val,
+                               uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeT2AddrModeImm8(MCInst &Inst, unsigned Val,
                                uint64_t Address, const void *Decoder);
+template<int shift>
+static DecodeStatus DecodeTAddrModeImm7(MCInst &Inst, unsigned Val,
+                               uint64_t Address, const void *Decoder);
+template<int shift, int WriteBack>
+static DecodeStatus DecodeT2AddrModeImm7(MCInst &Inst, unsigned Val,
+                               uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeThumbAddSPImm(MCInst &Inst, uint16_t Val,
                                uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeThumbAddSPReg(MCInst &Inst, uint16_t Insn,
                                 uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeThumbCPS(MCInst &Inst, uint16_t Insn,
                                 uint64_t Address, const void *Decoder);
 static DecodeStatus DecodeQADDInstruction(MCInst &Inst, unsigned Insn,
                                 uint64_t Address, const void *Decoder);

 static DecodeStatus DecodeRestrictedFPPredicateOperand(MCInst &Inst,
                                                        unsigned Val,
                                                        uint64_t Address,
                                                        const void *Decoder);
 template<bool Writeback>
 static DecodeStatus DecodeVSTRVLDR_SYSREG(MCInst &Inst, unsigned Insn,
                                           uint64_t Address,
                                           const void *Decoder);
+template<int shift>
+static DecodeStatus DecodeMVE_MEM_1_pre(MCInst &Inst, unsigned Val,
+                                        uint64_t Address, const void *Decoder);
+template<int shift>
+static DecodeStatus DecodeMVE_MEM_2_pre(MCInst &Inst, unsigned Val,
+                                        uint64_t Address, const void *Decoder);
+template<int shift>
+static DecodeStatus DecodeMVE_MEM_3_pre(MCInst &Inst, unsigned Val,
+                                        uint64_t Address, const void *Decoder);
 template<unsigned MinLog, unsigned MaxLog>
 static DecodeStatus DecodePowerTwoOperand(MCInst &Inst, unsigned Val,
                                           uint64_t Address,
                                           const void *Decoder);
 template <int shift>
 static DecodeStatus DecodeExpandedImmOperand(MCInst &Inst, unsigned Val,
                                              uint64_t Address,
                                              const void *Decoder);
if (Val == 0)
     imm = INT32_MIN;
   else if (!(Val & 0x100))
     imm *= -1;
   Inst.addOperand(MCOperand::createImm(imm));
 
   return MCDisassembler::Success;
 }
 
+template<int shift>
+static DecodeStatus DecodeT2Imm7(MCInst &Inst, unsigned Val,
+                         uint64_t Address, const void *Decoder) {
+  int imm = Val & 0x7F;
+  if (Val == 0)
+    imm = INT32_MIN;
+  else if (!(Val & 0x80))
+    imm *= -1;
+  if (imm != INT32_MIN)
+    imm <<= shift;
+  Inst.addOperand(MCOperand::createImm(imm));
+
+  return MCDisassembler::Success;
+}
+
 static DecodeStatus DecodeT2AddrModeImm8(MCInst &Inst, unsigned Val,
                                  uint64_t Address, const void *Decoder) {
   DecodeStatus S = MCDisassembler::Success;
 
   unsigned Rn = fieldFromInstruction(Val, 9, 4);
   unsigned imm = fieldFromInstruction(Val, 0, 9);
 
   // Thumb stores cannot use PC as dest register.
static DecodeStatus DecodeT2AddrModeImm8(MCInst &Inst, unsigned Val,
   if (!Check(S, DecodeGPRRegisterClass(Inst, Rn, Address, Decoder)))
     return MCDisassembler::Fail;
   if (!Check(S, DecodeT2Imm8(Inst, imm, Address, Decoder)))
     return MCDisassembler::Fail;
 
   return S;
 }
 
+template<int shift>
+static DecodeStatus DecodeTAddrModeImm7(MCInst &Inst, unsigned Val,
+                                         uint64_t Address,
+                                         const void *Decoder) {
+  DecodeStatus S = MCDisassembler::Success;
+
+  unsigned Rn = fieldFromInstruction(Val, 8, 3);
+  unsigned imm =
+      fieldFromInstruction(Val, 0, 8);
+
+  if (!Check(S, DecodetGPRRegisterClass(Inst, Rn, Address, Decoder)))
+    return MCDisassembler::Fail;
+  if (!Check(S, DecodeT2Imm7<shift>(Inst, imm, Address, Decoder)))
+    return MCDisassembler::Fail;
+
+  return S;
+}
+
+template<int shift, int WriteBack>
+static DecodeStatus DecodeT2AddrModeImm7(MCInst &Inst, unsigned Val,
+                                         uint64_t Address,
+                                         const void *Decoder) {
+  DecodeStatus S = MCDisassembler::Success;
+
+  unsigned Rn = fieldFromInstruction(Val, 8, 4);
+  unsigned imm = fieldFromInstruction(Val, 0, 8);
+  if (WriteBack) {
+    if (!Check(S, DecoderGPRRegisterClass(Inst, Rn, Address, Decoder)))
+      return MCDisassembler::Fail;
+  } else if (!Check(S, DecodeGPRnopcRegisterClass(Inst, Rn, Address, Decoder)))
+    return MCDisassembler::Fail;
+  if (!Check(S, DecodeT2Imm7<shift>(Inst, imm, Address, Decoder)))
+    return MCDisassembler::Fail;
+
+  return S;
+}
+
 static DecodeStatus DecodeT2LdStPre(MCInst &Inst, unsigned Insn,
                                     uint64_t Address, const void *Decoder) {
   DecodeStatus S = MCDisassembler::Success;
 
   unsigned Rt = fieldFromInstruction(Insn, 12, 4);
   unsigned Rn = fieldFromInstruction(Insn, 16, 4);
   unsigned addr = fieldFromInstruction(Insn, 0, 8);
   addr |= fieldFromInstruction(Insn, 9, 1) << 8;
static DecodeStatus DecodePostIdxReg(MCInst &Inst, unsigned Insn,
 
   if (!Check(S, DecodeGPRnopcRegisterClass(Inst, Rm, Address, Decoder)))
     return MCDisassembler::Fail;
   Inst.addOperand(MCOperand::createImm(add));
 
   return S;
 }
 
+static DecodeStatus DecodeMveAddrModeRQ(MCInst &Inst, unsigned Insn,
+                             uint64_t Address, const void *Decoder) {
+  DecodeStatus S = MCDisassembler::Success;
+  unsigned Rn = fieldFromInstruction(Insn, 3, 4);
+  unsigned Qm = fieldFromInstruction(Insn, 0, 3);
+
+  if (!Check(S, DecodeGPRnopcRegisterClass(Inst, Rn, Address, Decoder)))
+    return MCDisassembler::Fail;
+  if (!Check(S, DecodeMQPRRegisterClass(Inst, Qm, Address, Decoder)))
+    return MCDisassembler::Fail;
+
+  return S;
+}
+
+template<int shift>
+static DecodeStatus DecodeMveAddrModeQ(MCInst &Inst, unsigned Insn,
+                             uint64_t Address, const void *Decoder) {
+  DecodeStatus S = MCDisassembler::Success;
+  unsigned Qm = fieldFromInstruction(Insn, 8, 3);
+  int imm = fieldFromInstruction(Insn, 0, 7);
+
+  if (!Check(S, DecodeMQPRRegisterClass(Inst, Qm, Address, Decoder)))
+    return MCDisassembler::Fail;
+
+  if(!fieldFromInstruction(Insn, 7, 1)) {
+    if (imm == 0)
+      imm = INT32_MIN;                 // indicate -0
+    else
+      imm *= -1;
+  }
+  if (imm != INT32_MIN)
+    imm <<= shift;
+  Inst.addOperand(MCOperand::createImm(imm));
+
+  return S;
+}
+
 static DecodeStatus DecodeThumbBLXOffset(MCInst &Inst, unsigned Val,
                                  uint64_t Address, const void *Decoder) {
   // Val is passed in as S:J1:J2:imm10H:imm10L:'0'
   // Note only one trailing zero not two.  Also the J1 and J2 values are from
   // the encoded instruction.  So here change to I1 and I2 values via:
   // I1 = NOT(J1 EOR S);
   // I2 = NOT(J2 EOR S);
   // and build the imm32 with two trailing zeros as documented:
if (!Check(S, DecodeT2AddrModeImm7s4(Inst, addr, Address, Decoder)))
     return MCDisassembler::Fail;
 
   Inst.addOperand(MCOperand::createImm(ARMCC::AL));
   Inst.addOperand(MCOperand::createReg(0));
 
   return S;
 }
 
+static inline DecodeStatus DecodeMVE_MEM_pre(
+  MCInst &Inst, unsigned Val, uint64_t Address, const void *Decoder,
+  unsigned Rn, OperandDecoder RnDecoder, OperandDecoder AddrDecoder) {
+  DecodeStatus S = MCDisassembler::Success;
+
+  unsigned Qd = fieldFromInstruction(Val, 13, 3);
+  unsigned addr = fieldFromInstruction(Val, 0, 7) |
+                  (fieldFromInstruction(Val, 23, 1) << 7) | (Rn << 8);
+
+  if (!Check(S, RnDecoder(Inst, Rn, Address, Decoder)))
+    return MCDisassembler::Fail;
+  if (!Check(S, DecodeMQPRRegisterClass(Inst, Qd, Address, Decoder)))
+    return MCDisassembler::Fail;
+  if (!Check(S, AddrDecoder(Inst, addr, Address, Decoder)))
+    return MCDisassembler::Fail;
+
+  return S;
+}
+
+template <int shift>
+static DecodeStatus DecodeMVE_MEM_1_pre(MCInst &Inst, unsigned Val,
+                                        uint64_t Address, const void *Decoder) {
+  return DecodeMVE_MEM_pre(Inst, Val, Address, Decoder,
+                           fieldFromInstruction(Val, 16, 3),
+                           DecodetGPRRegisterClass,
+                           DecodeTAddrModeImm7<shift>);
+}
+
+template <int shift>
+static DecodeStatus DecodeMVE_MEM_2_pre(MCInst &Inst, unsigned Val,
+                                        uint64_t Address, const void *Decoder) {
+  return DecodeMVE_MEM_pre(Inst, Val, Address, Decoder,
+                           fieldFromInstruction(Val, 16, 4),
+                           DecoderGPRRegisterClass,
+                           DecodeT2AddrModeImm7<shift,1>);
+}
+
+template <int shift>
+static DecodeStatus DecodeMVE_MEM_3_pre(MCInst &Inst, unsigned Val,
+                                        uint64_t Address, const void *Decoder) {
+  return DecodeMVE_MEM_pre(Inst, Val, Address, Decoder,
+                           fieldFromInstruction(Val, 17, 3),
+                           DecodeMQPRRegisterClass,
+                           DecodeMveAddrModeQ<shift>);
+}
+
 template<unsigned MinLog, unsigned MaxLog>
 static DecodeStatus DecodePowerTwoOperand(MCInst &Inst, unsigned Val,
                                           uint64_t Address,
                                           const void *Decoder) {
   DecodeStatus S = MCDisassembler::Success;
 
   if (Val < MinLog || Val > MaxLog)
     return MCDisassembler::Fail;

llvm/lib/Target/ARM/MCTargetDesc/ARMAddressingModes.h

 /// ARM_AM - ARM Addressing Mode Stuff
 namespace ARM_AM {
   enum ShiftOpc {
     no_shift = 0,
     asr,
     lsl,
     lsr,
     ror,
-    rrx
+    rrx,
+    uxtw
   };
 
   enum AddrOpc {
     sub = 0,
     add
   };
 
   inline const char *getAddrOpcStr(AddrOpc Op) { return Op == sub ? "-" : ""; }
 
   inline const char *getShiftOpcStr(ShiftOpc Op) {
     switch (Op) {
     default: llvm_unreachable("Unknown shift opc!");
     case ARM_AM::asr: return "asr";
     case ARM_AM::lsl: return "lsl";
     case ARM_AM::lsr: return "lsr";
     case ARM_AM::ror: return "ror";
     case ARM_AM::rrx: return "rrx";
+    case ARM_AM::uxtw: return "uxtw";
     }
   }
 
   inline unsigned getShiftOpcEncoding(ShiftOpc Op) {
     switch (Op) {
     default: llvm_unreachable("Unknown shift opc!");
     case ARM_AM::asr: return 2;
     case ARM_AM::lsl: return 0;

llvm/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h

enum AddrMode {
     AddrModeT2_i8   = 12,
     AddrModeT2_so   = 13,
     AddrModeT2_pc   = 14, // +/- i12 for pc relative data
     AddrModeT2_i8s4 = 15, // i8 * 4
     AddrMode_i12    = 16,
     AddrMode5FP16   = 17,  // i8 * 2
     AddrModeT2_ldrex = 18, // i8 * 4, with unscaled offset in MCInst
     AddrModeT2_i7s4 = 19, // i7 * 4
+    AddrModeT2_i7s2 = 20, // i7 * 2
+    AddrModeT2_i7   = 21, // i7 * 1
   };
 
   inline static const char *AddrModeToString(AddrMode addrmode) {
     switch (addrmode) {
     case AddrModeNone:    return "AddrModeNone";
     case AddrMode1:       return "AddrMode1";
     case AddrMode2:       return "AddrMode2";
     case AddrMode3:       return "AddrMode3";
     case AddrMode4:       return "AddrMode4";
     case AddrMode5:       return "AddrMode5";
     case AddrMode5FP16:   return "AddrMode5FP16";
     case AddrMode6:       return "AddrMode6";
     case AddrModeT1_1:    return "AddrModeT1_1";
     case AddrModeT1_2:    return "AddrModeT1_2";
     case AddrModeT1_4:    return "AddrModeT1_4";
     case AddrModeT1_s:    return "AddrModeT1_s";
     case AddrModeT2_i12:  return "AddrModeT2_i12";
     case AddrModeT2_i8:   return "AddrModeT2_i8";
     case AddrModeT2_so:   return "AddrModeT2_so";
     case AddrModeT2_pc:   return "AddrModeT2_pc";
     case AddrModeT2_i8s4: return "AddrModeT2_i8s4";
     case AddrMode_i12:    return "AddrMode_i12";
     case AddrModeT2_ldrex:return "AddrModeT2_ldrex";
     case AddrModeT2_i7s4: return "AddrModeT2_i7s4";
+    case AddrModeT2_i7s2: return "AddrModeT2_i7s2";
+    case AddrModeT2_i7:   return "AddrModeT2_i7";
     }
   }
 
   /// Target Operand Flag enum.
   enum TOF {
     //===------------------------------------------------------------------===//
     // ARM Specific MachineOperand flags.
 

llvm/lib/Target/ARM/MCTargetDesc/ARMInstPrinter.h

public:
   void printComplexRotationOp(const MCInst *MI, unsigned OpNum,
                               const MCSubtargetInfo &STI, raw_ostream &O);
   // MVE
   void printVPTPredicateOperand(const MCInst *MI, unsigned OpNum,
                                 const MCSubtargetInfo &STI,
                                 raw_ostream &O);
   void printVPTMask(const MCInst *MI, unsigned OpNum,
                     const MCSubtargetInfo &STI, raw_ostream &O);
+  template<int shift>
+  void printMveAddrModeRQOperand(const MCInst *MI, unsigned OpNum,
+                                 const MCSubtargetInfo &STI, raw_ostream &O);
+  void printMveAddrModeQOperand(const MCInst *MI, unsigned OpNum,
+                                const MCSubtargetInfo &STI, raw_ostream &O);
   void printExpandedImmOperand(const MCInst *MI, unsigned OpNum,
                                const MCSubtargetInfo &STI, raw_ostream &O);
 
 private:
   unsigned DefaultAltIdx = ARM::NoRegAltName;
 };
 
 } // end namespace llvm
 
 #endif // LLVM_LIB_TARGET_ARM_MCTARGETDESC_ARMINSTPRINTER_H

llvm/lib/Target/ARM/MCTargetDesc/ARMInstPrinter.cpp

void ARMInstPrinter::printPostIdxImm8s4Operand(const MCInst *MI, unsigned OpNum,
                                                const MCSubtargetInfo &STI,
                                                raw_ostream &O) {
   const MCOperand &MO = MI->getOperand(OpNum);
   unsigned Imm = MO.getImm();
   O << markup("<imm:") << '#' << ((Imm & 256) ? "" : "-") << ((Imm & 0xff) << 2)
     << markup(">");
 }
 
+template<int shift>
+void ARMInstPrinter::printMveAddrModeRQOperand(const MCInst *MI, unsigned OpNum,
+                                               const MCSubtargetInfo &STI,
+                                               raw_ostream &O) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  const MCOperand &MO2 = MI->getOperand(OpNum + 1);
+
+  O << markup("<mem:") << "[";
+  printRegName(O, MO1.getReg());
+  O << ", ";
+  printRegName(O, MO2.getReg());
+
+  if (shift > 0)
+    printRegImmShift(O, ARM_AM::uxtw, shift, UseMarkup);
+
+  O << "]" << markup(">");
+}
+
+void ARMInstPrinter::printMveAddrModeQOperand(const MCInst *MI, unsigned OpNum,
+                                               const MCSubtargetInfo &STI,
+                                               raw_ostream &O) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  const MCOperand &MO2 = MI->getOperand(OpNum + 1);
+
+  O << markup("<mem:") << "[";
+  printRegName(O, MO1.getReg());
+
+  int64_t Imm = MO2.getImm();
+  if (Imm != 0)
+    O << ", " << markup("<imm:") << '#' << Imm << markup(">");
+
+  O << "]" << markup(">");
+}
+
 void ARMInstPrinter::printLdStmModeOperand(const MCInst *MI, unsigned OpNum,
                                            const MCSubtargetInfo &STI,
                                            raw_ostream &O) {
   ARM_AM::AMSubMode Mode =
       ARM_AM::getAM4SubMode(MI->getOperand(OpNum).getImm());
   O << ARM_AM::getAMSubModeStr(Mode);
 }
 

llvm/lib/Target/ARM/MCTargetDesc/ARMMCCodeEmitter.cpp

public:
 
   /// getT2ScaledImmOpValue - Return encoding info for '+/- immX<<Y'
   /// operand.
   template<unsigned Bits, unsigned Shift>
   uint32_t getT2ScaledImmOpValue(const MCInst &MI, unsigned OpIdx,
                                  SmallVectorImpl<MCFixup> &Fixups,
                                  const MCSubtargetInfo &STI) const;
 
+  /// getMveAddrModeRQOpValue - Return encoding info for 'reg, vreg'
+  /// operand.
+  uint32_t getMveAddrModeRQOpValue(const MCInst &MI, unsigned OpIdx,
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const;
+
+  /// getMveAddrModeQOpValue - Return encoding info for 'reg +/- imm7<<{shift}'
+  /// operand.
+  template<int shift>
+  uint32_t getMveAddrModeQOpValue(const MCInst &MI, unsigned OpIdx,
+                                  SmallVectorImpl<MCFixup> &Fixups,
+                                  const MCSubtargetInfo &STI) const;
+
   /// getLdStSORegOpValue - Return encoding info for 'reg +/- reg shop imm'
   /// operand as needed by load/store instructions.
   uint32_t getLdStSORegOpValue(const MCInst &MI, unsigned OpIdx,
                                SmallVectorImpl<MCFixup> &Fixups,
                                const MCSubtargetInfo &STI) const;
 
   /// getLdStmModeOpValue - Return encoding for load/store multiple mode.
   uint32_t getLdStmModeOpValue(const MCInst &MI, unsigned OpIdx,
public:
   unsigned getShiftOp(ARM_AM::ShiftOpc ShOpc) const {
     switch (ShOpc) {
     case ARM_AM::no_shift:
     case ARM_AM::lsl: return 0;
     case ARM_AM::lsr: return 1;
     case ARM_AM::asr: return 2;
     case ARM_AM::ror:
     case ARM_AM::rrx: return 3;
-    }
+    case ARM_AM::uxtw:
+    default:
       llvm_unreachable("Invalid ShiftOpc!");
     }
+  }
 
   /// getAddrMode2OffsetOpValue - Return encoding for am2offset operands.
   uint32_t getAddrMode2OffsetOpValue(const MCInst &MI, unsigned OpIdx,
                                      SmallVectorImpl<MCFixup> &Fixups,
                                      const MCSubtargetInfo &STI) const;
 
   /// getPostIdxRegOpValue - Return encoding for postidx_reg operands.
   uint32_t getPostIdxRegOpValue(const MCInst &MI, unsigned OpIdx,
unsigned getT2SOImmOpValue(const MCInst &MI, unsigned Op,
     unsigned Encoded =  ARM_AM::getT2SOImmVal(SoImm);
     assert(Encoded != ~0U && "Not a Thumb2 so_imm value?");
     return Encoded;
   }
 
   unsigned getT2AddrModeSORegOpValue(const MCInst &MI, unsigned OpNum,
     SmallVectorImpl<MCFixup> &Fixups,
     const MCSubtargetInfo &STI) const;
-  unsigned getT2AddrModeImm8OpValue(const MCInst &MI, unsigned OpNum,
+  template<unsigned Bits, unsigned Shift>
+  unsigned getT2AddrModeImmOpValue(const MCInst &MI, unsigned OpNum,
     SmallVectorImpl<MCFixup> &Fixups,
     const MCSubtargetInfo &STI) const;
   unsigned getT2AddrModeImm8OffsetOpValue(const MCInst &MI, unsigned OpNum,
     SmallVectorImpl<MCFixup> &Fixups,
     const MCSubtargetInfo &STI) const;
 
   /// getSORegOpValue - Return an encoded so_reg shifted register value.
   unsigned getSORegRegOpValue(const MCInst &MI, unsigned Op,
getT2ScaledImmOpValue(const MCInst &MI, unsigned OpIdx,
   Imm >>= Shift;
 
   uint32_t Binary = Imm & ((1U << Bits) - 1);
   // Immediate is always encoded as positive. The 'U' bit controls add vs sub.
   if (isAdd)
     Binary |= (1U << Bits);
   return Binary;
 }
+
+/// getMveAddrModeRQOpValue - Return encoding info for 'reg, vreg'
+/// operand.
+uint32_t ARMMCCodeEmitter::
+getMveAddrModeRQOpValue(const MCInst &MI, unsigned OpIdx,
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
+    // {6-3} Rn
+    // {2-0} Qm
+    const MCOperand &M0 = MI.getOperand(OpIdx);
+    const MCOperand &M1 = MI.getOperand(OpIdx + 1);
+
+    unsigned Rn = CTX.getRegisterInfo()->getEncodingValue(M0.getReg());
+    unsigned Qm = CTX.getRegisterInfo()->getEncodingValue(M1.getReg());
+
+    assert(Qm < 8 && "Qm is supposed to be encodable in 3 bits");
+
+    return (Rn << 3) | Qm;
+}
+
+/// getMveAddrModeRQOpValue - Return encoding info for 'reg, vreg'
+/// operand.
+template<int shift>
+uint32_t ARMMCCodeEmitter::
+getMveAddrModeQOpValue(const MCInst &MI, unsigned OpIdx,
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
+    // {10-8} Qm
+    // {7-0} Imm
+    const MCOperand &M0 = MI.getOperand(OpIdx);
+    const MCOperand &M1 = MI.getOperand(OpIdx + 1);
+
+    unsigned Qm = CTX.getRegisterInfo()->getEncodingValue(M0.getReg());
+    int32_t Imm = M1.getImm();
+
+    bool isAdd = Imm >= 0;
+
+    Imm >>= shift;
+
+    if (!isAdd)
+      Imm = -(uint32_t)Imm;
+
+    Imm &= 0x7f;
+
+    if (isAdd)
+      Imm |= 0x80;
+
+    assert(Qm < 8 && "Qm is supposed to be encodable in 3 bits");
+
+    return (Qm << 8) | Imm;
+}
+
 /// getT2AddrModeImm8s4OpValue - Return encoding info for
 /// 'reg +/- imm8<<2' operand.
 uint32_t ARMMCCodeEmitter::
 getT2AddrModeImm8s4OpValue(const MCInst &MI, unsigned OpIdx,
                         SmallVectorImpl<MCFixup> &Fixups,
                         const MCSubtargetInfo &STI) const {
   // {12-9} = reg
   // {8}    = (U)nsigned (add == '1', sub == '0')
getT2AddrModeSORegOpValue(const MCInst &MI, unsigned OpNum,
   Value <<= 4;
   Value |= CTX.getRegisterInfo()->getEncodingValue(MO2.getReg());
   Value <<= 2;
   Value |= MO3.getImm();
 
   return Value;
 }
 
+template<unsigned Bits, unsigned Shift>
 unsigned ARMMCCodeEmitter::
-getT2AddrModeImm8OpValue(const MCInst &MI, unsigned OpNum,
-                         SmallVectorImpl<MCFixup> &Fixups,
-                         const MCSubtargetInfo &STI) const {
+getT2AddrModeImmOpValue(const MCInst &MI, unsigned OpNum,
   const MCOperand &MO1 = MI.getOperand(OpNum);
   const MCOperand &MO2 = MI.getOperand(OpNum+1);
 
   // FIXME: Needs fixup support.
   unsigned Value = CTX.getRegisterInfo()->getEncodingValue(MO1.getReg());
 
-  // Even though the immediate is 8 bits long, we need 9 bits in order
+  // If the immediate is B bits long, we need B+1 bits in order
   // to represent the (inverse of the) sign bit.
-  Value <<= 9;
-  int32_t tmp = (int32_t)MO2.getImm();
+  Value <<= (Bits + 1);
+  int32_t tmp = (int32_t)MO2.getImm() >> Shift;
   if (tmp < 0)
     tmp = abs(tmp);
   else
-    Value |= 256; // Set the ADD bit
-  Value |= tmp & 255;
+    Value |= (1U << Bits); // Set the ADD bit
+  Value |= tmp & ((1U << Bits) - 1);
   return Value;
 }
 
 unsigned ARMMCCodeEmitter::
 getT2AddrModeImm8OffsetOpValue(const MCInst &MI, unsigned OpNum,
                          SmallVectorImpl<MCFixup> &Fixups,
                          const MCSubtargetInfo &STI) const {
   const MCOperand &MO1 = MI.getOperand(OpNum);

llvm/lib/Target/ARM/Thumb2InstrInfo.cpp

if (AddrMode == ARMII::AddrModeT2_i8 || AddrMode == ARMII::AddrModeT2_i12) {
       NumBits = 8;
       Scale = 2;
       Offset += InstrOffs * 2;
       assert((Offset & (Scale-1)) == 0 && "Can't encode this offset!");
       if (Offset < 0) {
         Offset = -Offset;
         isSub = true;
       }
-    } else if (AddrMode == ARMII::AddrModeT2_i7s4) {
+    } else if (AddrMode == ARMII::AddrModeT2_i7s4 ||
+               AddrMode == ARMII::AddrModeT2_i7s2 ||
+               AddrMode == ARMII::AddrModeT2_i7) {
       Offset += MI.getOperand(FrameRegIdx + 1).getImm();
-      NumBits = 9; // 7 bits scaled by 4
-      unsigned OffsetMask = 0x3;
+      NumBits = 7;
+      unsigned OffsetMask = 0x0;
+      switch (AddrMode) {
+      case ARMII::AddrModeT2_i7s4: NumBits += 2; OffsetMask = 0x3; break;
+      case ARMII::AddrModeT2_i7s2: NumBits += 1; OffsetMask = 0x1; break;
+      }
       // MCInst operand expects already scaled value.
       Scale = 1;
       assert((Offset & OffsetMask) == 0 && "Can't encode this offset!");
       (void)OffsetMask; // squash unused-variable warning at -NDEBUG
     } else if (AddrMode == ARMII::AddrModeT2_i8s4) {
       Offset += MI.getOperand(FrameRegIdx + 1).getImm() * 4;
-      NumBits = 10; // 8 bits scaled by 4
+      NumBits = 8 + 2;
       // MCInst operand expects already scaled value.
       Scale = 1;
       assert((Offset & 3) == 0 && "Can't encode this offset!");
     } else if (AddrMode == ARMII::AddrModeT2_ldrex) {
       Offset += MI.getOperand(FrameRegIdx + 1).getImm() * 4;
       NumBits = 8; // 8 bits scaled by 4
       Scale = 4;
       assert((Offset & 3) == 0 && "Can't encode this offset!");

llvm/test/MC/ARM/mve-load-store.s

+# RUN: not llvm-mc -triple=thumbv8.1m.main-none-eabi -mattr=+mve -show-encoding  < %s \
+# RUN:   | FileCheck --check-prefix=CHECK-NOFP %s
+# RUN: not llvm-mc -triple=thumbv8.1m.main-none-eabi -mattr=+mve.fp,+fp64 -show-encoding  < %s 2>%t \
+# RUN:   | FileCheck --check-prefix=CHECK %s
+# RUN:     FileCheck --check-prefix=ERROR < %t %s
+
+# CHECK: vldrb.u8 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1e]
+vldrb.u8 q0, [r0]
+
+# CHECK: vldrb.u8 q1, [r0] @ encoding: [0x90,0xed,0x00,0x3e]
+# CHECK-NOFP: vldrb.u8 q1, [r0] @ encoding: [0x90,0xed,0x00,0x3e]
+vldrb.u8 q1, [r0]
+
+# CHECK: vldrb.u8 q0, [r11] @ encoding: [0x9b,0xed,0x00,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r11] @ encoding: [0x9b,0xed,0x00,0x1e]
+vldrb.u8 q0, [r11]
+
+# CHECK: vldrb.u8 q3, [r11] @ encoding: [0x9b,0xed,0x00,0x7e]
+# CHECK-NOFP: vldrb.u8 q3, [r11] @ encoding: [0x9b,0xed,0x00,0x7e]
+vldrb.u8 q3, [r11]
+
+# CHECK: vldrb.u8 q0, [r4, #56] @ encoding: [0x94,0xed,0x38,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r4, #56] @ encoding: [0x94,0xed,0x38,0x1e]
+vldrb.u8 q0, [r4, #56]
+
+# CHECK: vldrb.u8 q4, [r4, #56] @ encoding: [0x94,0xed,0x38,0x9e]
+# CHECK-NOFP: vldrb.u8 q4, [r4, #56] @ encoding: [0x94,0xed,0x38,0x9e]
+vldrb.u8 q4, [r4, #56]
+
+# CHECK: vldrb.u8 q0, [r8, #56] @ encoding: [0x98,0xed,0x38,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r8, #56] @ encoding: [0x98,0xed,0x38,0x1e]
+vldrb.u8 q0, [r8, #56]
+
+# CHECK: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+vldrb.u8 q5, [r4, #56]!
+
+# CHECK: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+vldrb.u8 q5, [r4, #56]!
+
+# CHECK: vldrb.u8 q5, [r4], #-25 @ encoding: [0x34,0xec,0x19,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [r4], #-25 @ encoding: [0x34,0xec,0x19,0xbe]
+vldrb.u8 q5, [r4], #-25
+
+# CHECK: vldrb.u8 q5, [r10], #-25 @ encoding: [0x3a,0xec,0x19,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [r10], #-25 @ encoding: [0x3a,0xec,0x19,0xbe]
+vldrb.u8 q5, [r10], #-25
+
+# CHECK: vldrb.u8 q5, [sp, #-25] @ encoding: [0x1d,0xed,0x19,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [sp, #-25] @ encoding: [0x1d,0xed,0x19,0xbe]
+vldrb.u8 q5, [sp, #-25]
+
+# CHECK: vldrb.u8 q5, [sp, #-127] @ encoding: [0x1d,0xed,0x7f,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [sp, #-127] @ encoding: [0x1d,0xed,0x7f,0xbe]
+vldrb.u8 q5, [sp, #-127]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrb.u8 q0, [r0, #128]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrb.u8 q0, [r0, #-128]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrb.u8 q0, [r0], #128
+
+# CHECK: vstrb.8 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1e]
+# CHECK-NOFP: vstrb.8 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1e]
+vstrb.8 q0, [r0]
+
+# CHECK: vstrb.8 q1, [r0] @ encoding: [0x80,0xed,0x00,0x3e]
+# CHECK-NOFP: vstrb.8 q1, [r0] @ encoding: [0x80,0xed,0x00,0x3e]
+vstrb.8 q1, [r0]
+
+# CHECK: vstrb.8 q0, [r11] @ encoding: [0x8b,0xed,0x00,0x1e]
+# CHECK-NOFP: vstrb.8 q0, [r11] @ encoding: [0x8b,0xed,0x00,0x1e]
+vstrb.8 q0, [r11]
+
+# CHECK: vstrb.8 q3, [r11] @ encoding: [0x8b,0xed,0x00,0x7e]
+# CHECK-NOFP: vstrb.8 q3, [r11] @ encoding: [0x8b,0xed,0x00,0x7e]
+vstrb.8 q3, [r11]
+
+# CHECK: vstrb.8 q0, [r4, #56] @ encoding: [0x84,0xed,0x38,0x1e]
+# CHECK-NOFP: vstrb.8 q0, [r4, #56] @ encoding: [0x84,0xed,0x38,0x1e]
+vstrb.8 q0, [r4, #56]
+
+# CHECK: vstrb.8 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x9e]
+# CHECK-NOFP: vstrb.8 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x9e]
+vstrb.8 q4, [r4, #56]
+
+# CHECK: vstrb.8 q0, [r8, #56] @ encoding: [0x88,0xed,0x38,0x1e]
+# CHECK-NOFP: vstrb.8 q0, [r8, #56] @ encoding: [0x88,0xed,0x38,0x1e]
+vstrb.8 q0, [r8, #56]
+
+# CHECK: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+vstrb.8 q5, [r4, #56]!
+
+# CHECK: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+vstrb.8 q5, [r4, #56]!
+
+# CHECK: vstrb.8 q5, [r4], #-25 @ encoding: [0x24,0xec,0x19,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [r4], #-25 @ encoding: [0x24,0xec,0x19,0xbe]
+vstrb.8 q5, [r4], #-25
+
+# CHECK: vstrb.8 q5, [r10], #-25 @ encoding: [0x2a,0xec,0x19,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [r10], #-25 @ encoding: [0x2a,0xec,0x19,0xbe]
+vstrb.8 q5, [r10], #-25
+
+# CHECK: vstrb.8 q5, [sp, #-25] @ encoding: [0x0d,0xed,0x19,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [sp, #-25] @ encoding: [0x0d,0xed,0x19,0xbe]
+vstrb.8 q5, [sp, #-25]
+
+# CHECK: vstrb.8 q5, [sp, #127] @ encoding: [0x8d,0xed,0x7f,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [sp, #127] @ encoding: [0x8d,0xed,0x7f,0xbe]
+vstrb.8 q5, [sp, #127]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrb.u8 q0, [r0, #128]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrb.u8 q0, [r0, #-128]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrb.u8 q0, [r0], #128
+
+# CHECK: vldrb.u16 q0, [r0] @ encoding: [0x90,0xfd,0x80,0x0e]
+# CHECK-NOFP: vldrb.u16 q0, [r0] @ encoding: [0x90,0xfd,0x80,0x0e]
+vldrb.u16 q0, [r0]
+
+# CHECK: vldrb.u16 q1, [r0] @ encoding: [0x90,0xfd,0x80,0x2e]
+# CHECK-NOFP: vldrb.u16 q1, [r0] @ encoding: [0x90,0xfd,0x80,0x2e]
+vldrb.u16 q1, [r0]
+
+# CHECK: vldrb.u16 q0, [r7] @ encoding: [0x97,0xfd,0x80,0x0e]
+# CHECK-NOFP: vldrb.u16 q0, [r7] @ encoding: [0x97,0xfd,0x80,0x0e]
+vldrb.u16 q0, [r7]
+
+# CHECK: vldrb.u16 q3, [r7] @ encoding: [0x97,0xfd,0x80,0x6e]
+# CHECK-NOFP: vldrb.u16 q3, [r7] @ encoding: [0x97,0xfd,0x80,0x6e]
+vldrb.u16 q3, [r7]
+
+# CHECK: vldrb.u16 q0, [r4, #56] @ encoding: [0x94,0xfd,0xb8,0x0e]
+# CHECK-NOFP: vldrb.u16 q0, [r4, #56] @ encoding: [0x94,0xfd,0xb8,0x0e]
+vldrb.u16 q0, [r4, #56]
+
+# CHECK: vldrb.u16 q4, [r4, #56] @ encoding: [0x94,0xfd,0xb8,0x8e]
+# CHECK-NOFP: vldrb.u16 q4, [r4, #56] @ encoding: [0x94,0xfd,0xb8,0x8e]
+vldrb.u16 q4, [r4, #56]
+
+# CHECK: vldrb.u16 q0, [r2, #56] @ encoding: [0x92,0xfd,0xb8,0x0e]
+# CHECK-NOFP: vldrb.u16 q0, [r2, #56] @ encoding: [0x92,0xfd,0xb8,0x0e]
+vldrb.u16 q0, [r2, #56]
+
+# CHECK: vldrb.u16 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0xb8,0xae]
+# CHECK-NOFP: vldrb.u16 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0xb8,0xae]
+vldrb.u16 q5, [r4, #56]!
+
+# CHECK: vldrb.u16 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0xb8,0xae]
+# CHECK-NOFP: vldrb.u16 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0xb8,0xae]
+vldrb.u16 q5, [r4, #56]!
+
+# CHECK: vldrb.u16 q5, [r4], #-1 @ encoding: [0x34,0xfc,0x81,0xae]
+# CHECK-NOFP: vldrb.u16 q5, [r4], #-1 @ encoding: [0x34,0xfc,0x81,0xae]
+vldrb.u16 q5, [r4], #-1
+
+# CHECK: vldrb.u16 q5, [r3], #-25 @ encoding: [0x33,0xfc,0x99,0xae]
+# CHECK-NOFP: vldrb.u16 q5, [r3], #-25 @ encoding: [0x33,0xfc,0x99,0xae]
+vldrb.u16 q5, [r3], #-25
+
+# CHECK: vldrb.u16 q5, [r6, #-25] @ encoding: [0x16,0xfd,0x99,0xae]
+# CHECK-NOFP: vldrb.u16 q5, [r6, #-25] @ encoding: [0x16,0xfd,0x99,0xae]
+vldrb.u16 q5, [r6, #-25]
+
+# CHECK: vldrb.u16 q5, [r6, #-64] @ encoding: [0x16,0xfd,0xc0,0xae]
+# CHECK-NOFP: vldrb.u16 q5, [r6, #-64] @ encoding: [0x16,0xfd,0xc0,0xae]
+vldrb.u16 q5, [r6, #-64]
+
+# CHECK: vldrb.s16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x0e]
+# CHECK-NOFP: vldrb.s16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x0e]
+vldrb.s16 q0, [r0]
+
+# CHECK: vldrb.s16 q1, [r0] @ encoding: [0x90,0xed,0x80,0x2e]
+# CHECK-NOFP: vldrb.s16 q1, [r0] @ encoding: [0x90,0xed,0x80,0x2e]
+vldrb.s16 q1, [r0]
+
+# CHECK: vldrb.s16 q0, [r7] @ encoding: [0x97,0xed,0x80,0x0e]
+# CHECK-NOFP: vldrb.s16 q0, [r7] @ encoding: [0x97,0xed,0x80,0x0e]
+vldrb.s16 q0, [r7]
+
+# CHECK: vldrb.s16 q3, [r7] @ encoding: [0x97,0xed,0x80,0x6e]
+# CHECK-NOFP: vldrb.s16 q3, [r7] @ encoding: [0x97,0xed,0x80,0x6e]
+vldrb.s16 q3, [r7]
+
+# CHECK: vldrb.s16 q0, [r4, #56] @ encoding: [0x94,0xed,0xb8,0x0e]
+# CHECK-NOFP: vldrb.s16 q0, [r4, #56] @ encoding: [0x94,0xed,0xb8,0x0e]
+vldrb.s16 q0, [r4, #56]
+
+# CHECK: vldrb.s16 q4, [r4, #56] @ encoding: [0x94,0xed,0xb8,0x8e]
+# CHECK-NOFP: vldrb.s16 q4, [r4, #56] @ encoding: [0x94,0xed,0xb8,0x8e]
+vldrb.s16 q4, [r4, #56]
+
+# CHECK: vldrb.s16 q0, [r2, #56] @ encoding: [0x92,0xed,0xb8,0x0e]
+# CHECK-NOFP: vldrb.s16 q0, [r2, #56] @ encoding: [0x92,0xed,0xb8,0x0e]
+vldrb.s16 q0, [r2, #56]
+
+# CHECK: vldrb.s16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0xb8,0xae]
+# CHECK-NOFP: vldrb.s16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0xb8,0xae]
+vldrb.s16 q5, [r4, #56]!
+
+# CHECK: vldrb.s16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0xb8,0xae]
+# CHECK-NOFP: vldrb.s16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0xb8,0xae]
+vldrb.s16 q5, [r4, #56]!
+
+# CHECK: vldrb.s16 q5, [r4], #-25 @ encoding: [0x34,0xec,0x99,0xae]
+# CHECK-NOFP: vldrb.s16 q5, [r4], #-25 @ encoding: [0x34,0xec,0x99,0xae]
+vldrb.s16 q5, [r4], #-25
+
+# CHECK: vldrb.s16 q5, [r3], #-25 @ encoding: [0x33,0xec,0x99,0xae]
+# CHECK-NOFP: vldrb.s16 q5, [r3], #-25 @ encoding: [0x33,0xec,0x99,0xae]
+vldrb.s16 q5, [r3], #-25
+
+# CHECK: vldrb.s16 q5, [r6, #-25] @ encoding: [0x16,0xed,0x99,0xae]
+# CHECK-NOFP: vldrb.s16 q5, [r6, #-25] @ encoding: [0x16,0xed,0x99,0xae]
+vldrb.s16 q5, [r6, #-25]
+
+# CHECK: vldrb.s16 q5, [r6, #-64] @ encoding: [0x16,0xed,0xc0,0xae]
+# CHECK-NOFP: vldrb.s16 q5, [r6, #-64] @ encoding: [0x16,0xed,0xc0,0xae]
+vldrb.s16 q5, [r6, #-64]
+
+# CHECK: vstrb.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x0e]
+# CHECK-NOFP: vstrb.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x0e]
+vstrb.16 q0, [r0]
+
+# CHECK: vstrb.16 q1, [r0] @ encoding: [0x80,0xed,0x80,0x2e]
+# CHECK-NOFP: vstrb.16 q1, [r0] @ encoding: [0x80,0xed,0x80,0x2e]
+vstrb.16 q1, [r0]
+
+# CHECK: vstrb.16 q0, [r7] @ encoding: [0x87,0xed,0x80,0x0e]
+# CHECK-NOFP: vstrb.16 q0, [r7] @ encoding: [0x87,0xed,0x80,0x0e]
+vstrb.16 q0, [r7]
+
+# CHECK: vstrb.16 q3, [r7] @ encoding: [0x87,0xed,0x80,0x6e]
+# CHECK-NOFP: vstrb.16 q3, [r7] @ encoding: [0x87,0xed,0x80,0x6e]
+vstrb.16 q3, [r7]
+
+# CHECK: vstrb.16 q0, [r4, #56] @ encoding: [0x84,0xed,0xb8,0x0e]
+# CHECK-NOFP: vstrb.16 q0, [r4, #56] @ encoding: [0x84,0xed,0xb8,0x0e]
+vstrb.16 q0, [r4, #56]
+
+# CHECK: vstrb.16 q4, [r4, #56] @ encoding: [0x84,0xed,0xb8,0x8e]
+# CHECK-NOFP: vstrb.16 q4, [r4, #56] @ encoding: [0x84,0xed,0xb8,0x8e]
+vstrb.16 q4, [r4, #56]
+
+# CHECK: vstrb.16 q0, [r5, #56] @ encoding: [0x85,0xed,0xb8,0x0e]
+# CHECK-NOFP: vstrb.16 q0, [r5, #56] @ encoding: [0x85,0xed,0xb8,0x0e]
+vstrb.16 q0, [r5, #56]
+
+# CHECK: vstrb.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0xb8,0xae]
+# CHECK-NOFP: vstrb.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0xb8,0xae]
+vstrb.16 q5, [r4, #56]!
+
+# CHECK: vstrb.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0xb8,0xae]
+# CHECK-NOFP: vstrb.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0xb8,0xae]
+vstrb.16 q5, [r4, #56]!
+
+# CHECK: vstrb.16 q5, [r4], #-25 @ encoding: [0x24,0xec,0x99,0xae]
+# CHECK-NOFP: vstrb.16 q5, [r4], #-25 @ encoding: [0x24,0xec,0x99,0xae]
+vstrb.16 q5, [r4], #-25
+
+# CHECK: vstrb.16 q5, [r3], #-25 @ encoding: [0x23,0xec,0x99,0xae]
+# CHECK-NOFP: vstrb.16 q5, [r3], #-25 @ encoding: [0x23,0xec,0x99,0xae]
+vstrb.16 q5, [r3], #-25
+
+# CHECK: vstrb.16 q5, [r2, #-25] @ encoding: [0x02,0xed,0x99,0xae]
+# CHECK-NOFP: vstrb.16 q5, [r2, #-25] @ encoding: [0x02,0xed,0x99,0xae]
+vstrb.16 q5, [r2, #-25]
+
+# CHECK: vstrb.16 q5, [r2, #-64] @ encoding: [0x02,0xed,0xc0,0xae]
+# CHECK-NOFP: vstrb.16 q5, [r2, #-64] @ encoding: [0x02,0xed,0xc0,0xae]
+vstrb.16 q5, [r2, #-64]
+
+# CHECK: vldrb.u32 q0, [r0] @ encoding: [0x90,0xfd,0x00,0x0f]
+# CHECK-NOFP: vldrb.u32 q0, [r0] @ encoding: [0x90,0xfd,0x00,0x0f]
+vldrb.u32 q0, [r0]
+
+# CHECK: vldrb.u32 q1, [r0] @ encoding: [0x90,0xfd,0x00,0x2f]
+# CHECK-NOFP: vldrb.u32 q1, [r0] @ encoding: [0x90,0xfd,0x00,0x2f]
+vldrb.u32 q1, [r0]
+
+# CHECK: vldrb.u32 q0, [r7] @ encoding: [0x97,0xfd,0x00,0x0f]
+# CHECK-NOFP: vldrb.u32 q0, [r7] @ encoding: [0x97,0xfd,0x00,0x0f]
+vldrb.u32 q0, [r7]
+
+# CHECK: vldrb.u32 q3, [r7] @ encoding: [0x97,0xfd,0x00,0x6f]
+# CHECK-NOFP: vldrb.u32 q3, [r7] @ encoding: [0x97,0xfd,0x00,0x6f]
+vldrb.u32 q3, [r7]
+
+# CHECK: vldrb.u32 q0, [r4, #56] @ encoding: [0x94,0xfd,0x38,0x0f]
+# CHECK-NOFP: vldrb.u32 q0, [r4, #56] @ encoding: [0x94,0xfd,0x38,0x0f]
+vldrb.u32 q0, [r4, #56]
+
+# CHECK: vldrb.u32 q4, [r4, #56] @ encoding: [0x94,0xfd,0x38,0x8f]
+# CHECK-NOFP: vldrb.u32 q4, [r4, #56] @ encoding: [0x94,0xfd,0x38,0x8f]
+vldrb.u32 q4, [r4, #56]
+
+# CHECK: vldrb.u32 q0, [r2, #56] @ encoding: [0x92,0xfd,0x38,0x0f]
+# CHECK-NOFP: vldrb.u32 q0, [r2, #56] @ encoding: [0x92,0xfd,0x38,0x0f]
+vldrb.u32 q0, [r2, #56]
+
+# CHECK: vldrb.u32 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0x38,0xaf]
+# CHECK-NOFP: vldrb.u32 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0x38,0xaf]
+vldrb.u32 q5, [r4, #56]!
+
+# CHECK: vldrb.u32 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0x38,0xaf]
+# CHECK-NOFP: vldrb.u32 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0x38,0xaf]
+vldrb.u32 q5, [r4, #56]!
+
+# CHECK: vldrb.u32 q5, [r4], #-25 @ encoding: [0x34,0xfc,0x19,0xaf]
+# CHECK-NOFP: vldrb.u32 q5, [r4], #-25 @ encoding: [0x34,0xfc,0x19,0xaf]
+vldrb.u32 q5, [r4], #-25
+
+# CHECK: vldrb.u32 q5, [r3], #-25 @ encoding: [0x33,0xfc,0x19,0xaf]
+# CHECK-NOFP: vldrb.u32 q5, [r3], #-25 @ encoding: [0x33,0xfc,0x19,0xaf]
+vldrb.u32 q5, [r3], #-25
+
+# CHECK: vldrb.u32 q5, [r6, #-25] @ encoding: [0x16,0xfd,0x19,0xaf]
+# CHECK-NOFP: vldrb.u32 q5, [r6, #-25] @ encoding: [0x16,0xfd,0x19,0xaf]
+vldrb.u32 q5, [r6, #-25]
+
+# CHECK: vldrb.u32 q5, [r6, #-64] @ encoding: [0x16,0xfd,0x40,0xaf]
+# CHECK-NOFP: vldrb.u32 q5, [r6, #-64] @ encoding: [0x16,0xfd,0x40,0xaf]
+vldrb.u32 q5, [r6, #-64]
+
+# CHECK: vldrb.s32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x0f]
+# CHECK-NOFP: vldrb.s32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x0f]
+vldrb.s32 q0, [r0]
+
+# CHECK: vldrb.s32 q1, [r0] @ encoding: [0x90,0xed,0x00,0x2f]
+# CHECK-NOFP: vldrb.s32 q1, [r0] @ encoding: [0x90,0xed,0x00,0x2f]
+vldrb.s32 q1, [r0]
+
+# CHECK: vldrb.s32 q0, [r7] @ encoding: [0x97,0xed,0x00,0x0f]
+# CHECK-NOFP: vldrb.s32 q0, [r7] @ encoding: [0x97,0xed,0x00,0x0f]
+vldrb.s32 q0, [r7]
+
+# CHECK: vldrb.s32 q3, [r7] @ encoding: [0x97,0xed,0x00,0x6f]
+# CHECK-NOFP: vldrb.s32 q3, [r7] @ encoding: [0x97,0xed,0x00,0x6f]
+vldrb.s32 q3, [r7]
+
+# CHECK: vldrb.s32 q0, [r4, #56] @ encoding: [0x94,0xed,0x38,0x0f]
+# CHECK-NOFP: vldrb.s32 q0, [r4, #56] @ encoding: [0x94,0xed,0x38,0x0f]
+vldrb.s32 q0, [r4, #56]
+
+# CHECK: vldrb.s32 q4, [r4, #56] @ encoding: [0x94,0xed,0x38,0x8f]
+# CHECK-NOFP: vldrb.s32 q4, [r4, #56] @ encoding: [0x94,0xed,0x38,0x8f]
+vldrb.s32 q4, [r4, #56]
+
+# CHECK: vldrb.s32 q0, [r2, #56] @ encoding: [0x92,0xed,0x38,0x0f]
+# CHECK-NOFP: vldrb.s32 q0, [r2, #56] @ encoding: [0x92,0xed,0x38,0x0f]
+vldrb.s32 q0, [r2, #56]
+
+# CHECK: vldrb.s32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xaf]
+# CHECK-NOFP: vldrb.s32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xaf]
+vldrb.s32 q5, [r4, #56]!
+
+# CHECK: vldrb.s32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xaf]
+# CHECK-NOFP: vldrb.s32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xaf]
+vldrb.s32 q5, [r4, #56]!
+
+# CHECK: vldrb.s32 q5, [r4], #-25 @ encoding: [0x34,0xec,0x19,0xaf]
+# CHECK-NOFP: vldrb.s32 q5, [r4], #-25 @ encoding: [0x34,0xec,0x19,0xaf]
+vldrb.s32 q5, [r4], #-25
+
+# CHECK: vldrb.s32 q5, [r3], #-25 @ encoding: [0x33,0xec,0x19,0xaf]
+# CHECK-NOFP: vldrb.s32 q5, [r3], #-25 @ encoding: [0x33,0xec,0x19,0xaf]
+vldrb.s32 q5, [r3], #-25
+
+# CHECK: vldrb.s32 q5, [r6, #-25] @ encoding: [0x16,0xed,0x19,0xaf]
+# CHECK-NOFP: vldrb.s32 q5, [r6, #-25] @ encoding: [0x16,0xed,0x19,0xaf]
+vldrb.s32 q5, [r6, #-25]
+
+# CHECK: vldrb.s32 q5, [r6, #-64] @ encoding: [0x16,0xed,0x40,0xaf]
+# CHECK-NOFP: vldrb.s32 q5, [r6, #-64] @ encoding: [0x16,0xed,0x40,0xaf]
+vldrb.s32 q5, [r6, #-64]
+
+# CHECK: vstrb.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x0f]
+# CHECK-NOFP: vstrb.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x0f]
+vstrb.32 q0, [r0]
+
+# CHECK: vstrb.32 q1, [r0] @ encoding: [0x80,0xed,0x00,0x2f]
+# CHECK-NOFP: vstrb.32 q1, [r0] @ encoding: [0x80,0xed,0x00,0x2f]
+vstrb.32 q1, [r0]
+
+# CHECK: vstrb.32 q0, [r7] @ encoding: [0x87,0xed,0x00,0x0f]
+# CHECK-NOFP: vstrb.32 q0, [r7] @ encoding: [0x87,0xed,0x00,0x0f]
+vstrb.32 q0, [r7]
+
+# CHECK: vstrb.32 q3, [r7] @ encoding: [0x87,0xed,0x00,0x6f]
+# CHECK-NOFP: vstrb.32 q3, [r7] @ encoding: [0x87,0xed,0x00,0x6f]
+vstrb.32 q3, [r7]
+
+# CHECK: vstrb.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x38,0x0f]
+# CHECK-NOFP: vstrb.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x38,0x0f]
+vstrb.32 q0, [r4, #56]
+
+# CHECK: vstrb.32 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x8f]
+# CHECK-NOFP: vstrb.32 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x8f]
+vstrb.32 q4, [r4, #56]
+
+# CHECK: vstrb.32 q0, [r5, #56] @ encoding: [0x85,0xed,0x38,0x0f]
+# CHECK-NOFP: vstrb.32 q0, [r5, #56] @ encoding: [0x85,0xed,0x38,0x0f]
+vstrb.32 q0, [r5, #56]
+
+# CHECK: vstrb.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xaf]
+# CHECK-NOFP: vstrb.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xaf]
+vstrb.32 q5, [r4, #56]!
+
+# CHECK: vstrb.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xaf]
+# CHECK-NOFP: vstrb.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xaf]
+vstrb.32 q5, [r4, #56]!
+
+# CHECK: vstrb.32 q5, [r4], #-25 @ encoding: [0x24,0xec,0x19,0xaf]
+# CHECK-NOFP: vstrb.32 q5, [r4], #-25 @ encoding: [0x24,0xec,0x19,0xaf]
+vstrb.32 q5, [r4], #-25
+
+# CHECK: vstrb.32 q5, [r3], #-25 @ encoding: [0x23,0xec,0x19,0xaf]
+# CHECK-NOFP: vstrb.32 q5, [r3], #-25 @ encoding: [0x23,0xec,0x19,0xaf]
+vstrb.32 q5, [r3], #-25
+
+# CHECK: vstrb.32 q5, [r2, #-25] @ encoding: [0x02,0xed,0x19,0xaf]
+# CHECK-NOFP: vstrb.32 q5, [r2, #-25] @ encoding: [0x02,0xed,0x19,0xaf]
+vstrb.32 q5, [r2, #-25]
+
+# CHECK: vstrb.32 q5, [r2, #-64] @ encoding: [0x02,0xed,0x40,0xaf]
+# CHECK-NOFP: vstrb.32 q5, [r2, #-64] @ encoding: [0x02,0xed,0x40,0xaf]
+vstrb.32 q5, [r2, #-64]
+
+# CHECK: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+vldrh.u16 q0, [r0]
+
+# CHECK: vldrh.u16 q1, [r0] @ encoding: [0x90,0xed,0x80,0x3e]
+# CHECK-NOFP: vldrh.u16 q1, [r0] @ encoding: [0x90,0xed,0x80,0x3e]
+vldrh.u16 q1, [r0]
+
+# CHECK: vldrh.u16 q0, [r11] @ encoding: [0x9b,0xed,0x80,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r11] @ encoding: [0x9b,0xed,0x80,0x1e]
+vldrh.u16 q0, [r11]
+
+# CHECK: vldrh.u16 q3, [r11] @ encoding: [0x9b,0xed,0x80,0x7e]
+# CHECK-NOFP: vldrh.u16 q3, [r11] @ encoding: [0x9b,0xed,0x80,0x7e]
+vldrh.u16 q3, [r11]
+
+# CHECK: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+vldrh.u16 q0, [r4, #56]
+
+# CHECK: vldrh.u16 q4, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x9e]
+# CHECK-NOFP: vldrh.u16 q4, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x9e]
+vldrh.u16 q4, [r4, #56]
+
+# CHECK: vldrh.u16 q0, [r8, #56] @ encoding: [0x98,0xed,0x9c,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r8, #56] @ encoding: [0x98,0xed,0x9c,0x1e]
+vldrh.u16 q0, [r8, #56]
+
+# CHECK: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+vldrh.u16 q5, [r4, #56]!
+
+# CHECK: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+vldrh.u16 q5, [r4, #56]!
+
+# CHECK: vldrh.u16 q5, [r4], #-26 @ encoding: [0x34,0xec,0x8d,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r4], #-26 @ encoding: [0x34,0xec,0x8d,0xbe]
+vldrh.u16 q5, [r4], #-26
+
+# CHECK: vldrh.u16 q5, [r10], #-26 @ encoding: [0x3a,0xec,0x8d,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r10], #-26 @ encoding: [0x3a,0xec,0x8d,0xbe]
+vldrh.u16 q5, [r10], #-26
+
+# CHECK: vldrh.u16 q5, [sp, #-26] @ encoding: [0x1d,0xed,0x8d,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [sp, #-26] @ encoding: [0x1d,0xed,0x8d,0xbe]
+vldrh.u16 q5, [sp, #-26]
+
+# CHECK: vldrh.u16 q5, [sp, #-64] @ encoding: [0x1d,0xed,0xa0,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [sp, #-64] @ encoding: [0x1d,0xed,0xa0,0xbe]
+vldrh.u16 q5, [sp, #-64]
+
+# CHECK: vldrh.u16 q5, [sp, #-254] @ encoding: [0x1d,0xed,0xff,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [sp, #-254] @ encoding: [0x1d,0xed,0xff,0xbe]
+vldrh.u16 q5, [sp, #-254]
+
+# CHECK: vldrh.u16 q5, [r10], #254 @ encoding: [0xba,0xec,0xff,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r10], #254 @ encoding: [0xba,0xec,0xff,0xbe]
+vldrh.u16 q5, [r10], #254
+
+# CHECK: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+vstrh.16 q0, [r0]
+
+# CHECK: vstrh.16 q1, [r0] @ encoding: [0x80,0xed,0x80,0x3e]
+# CHECK-NOFP: vstrh.16 q1, [r0] @ encoding: [0x80,0xed,0x80,0x3e]
+vstrh.16 q1, [r0]
+
+# CHECK: vstrh.16 q0, [r11] @ encoding: [0x8b,0xed,0x80,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r11] @ encoding: [0x8b,0xed,0x80,0x1e]
+vstrh.16 q0, [r11]
+
+# CHECK: vstrh.16 q3, [r11] @ encoding: [0x8b,0xed,0x80,0x7e]
+# CHECK-NOFP: vstrh.16 q3, [r11] @ encoding: [0x8b,0xed,0x80,0x7e]
+vstrh.16 q3, [r11]
+
+# CHECK: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+vstrh.16 q0, [r4, #56]
+
+# CHECK: vstrh.16 q4, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x9e]
+# CHECK-NOFP: vstrh.16 q4, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x9e]
+vstrh.16 q4, [r4, #56]
+
+# CHECK: vstrh.16 q0, [r8, #56] @ encoding: [0x88,0xed,0x9c,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r8, #56] @ encoding: [0x88,0xed,0x9c,0x1e]
+vstrh.16 q0, [r8, #56]
+
+# CHECK: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+vstrh.16 q5, [r4, #56]!
+
+# CHECK: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+vstrh.16 q5, [r4, #56]!
+
+# CHECK: vstrh.16 q5, [r4], #-26 @ encoding: [0x24,0xec,0x8d,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r4], #-26 @ encoding: [0x24,0xec,0x8d,0xbe]
+vstrh.16 q5, [r4], #-26
+
+# CHECK: vstrh.16 q5, [r10], #-26 @ encoding: [0x2a,0xec,0x8d,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r10], #-26 @ encoding: [0x2a,0xec,0x8d,0xbe]
+vstrh.16 q5, [r10], #-26
+
+# CHECK: vstrh.16 q5, [sp, #-26] @ encoding: [0x0d,0xed,0x8d,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [sp, #-26] @ encoding: [0x0d,0xed,0x8d,0xbe]
+vstrh.16 q5, [sp, #-26]
+
+# CHECK: vstrh.16 q5, [sp, #-64] @ encoding: [0x0d,0xed,0xa0,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [sp, #-64] @ encoding: [0x0d,0xed,0xa0,0xbe]
+vstrh.16 q5, [sp, #-64]
+
+# CHECK: vstrh.16 q5, [sp, #-254] @ encoding: [0x0d,0xed,0xff,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [sp, #-254] @ encoding: [0x0d,0xed,0xff,0xbe]
+vstrh.16 q5, [sp, #-254]
+
+# CHECK: vstrh.16 q5, [r10], #254 @ encoding: [0xaa,0xec,0xff,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r10], #254 @ encoding: [0xaa,0xec,0xff,0xbe]
+vstrh.16 q5, [r10], #254
+
+# CHECK: vldrh.u32 q0, [r0] @ encoding: [0x98,0xfd,0x00,0x0f]
+# CHECK-NOFP: vldrh.u32 q0, [r0] @ encoding: [0x98,0xfd,0x00,0x0f]
+vldrh.u32 q0, [r0]
+
+# CHECK: vldrh.u32 q1, [r0] @ encoding: [0x98,0xfd,0x00,0x2f]
+# CHECK-NOFP: vldrh.u32 q1, [r0] @ encoding: [0x98,0xfd,0x00,0x2f]
+vldrh.u32 q1, [r0]
+
+# CHECK: vldrh.u32 q0, [r7] @ encoding: [0x9f,0xfd,0x00,0x0f]
+# CHECK-NOFP: vldrh.u32 q0, [r7] @ encoding: [0x9f,0xfd,0x00,0x0f]
+vldrh.u32 q0, [r7]
+
+# CHECK: vldrh.u32 q3, [r7] @ encoding: [0x9f,0xfd,0x00,0x6f]
+# CHECK-NOFP: vldrh.u32 q3, [r7] @ encoding: [0x9f,0xfd,0x00,0x6f]
+vldrh.u32 q3, [r7]
+
+# CHECK: vldrh.u32 q0, [r4, #56] @ encoding: [0x9c,0xfd,0x1c,0x0f]
+# CHECK-NOFP: vldrh.u32 q0, [r4, #56] @ encoding: [0x9c,0xfd,0x1c,0x0f]
+vldrh.u32 q0, [r4, #56]
+
+# CHECK: vldrh.u32 q4, [r4, #56] @ encoding: [0x9c,0xfd,0x1c,0x8f]
+# CHECK-NOFP: vldrh.u32 q4, [r4, #56] @ encoding: [0x9c,0xfd,0x1c,0x8f]
+vldrh.u32 q4, [r4, #56]
+
+# CHECK: vldrh.u32 q0, [r2, #56] @ encoding: [0x9a,0xfd,0x1c,0x0f]
+# CHECK-NOFP: vldrh.u32 q0, [r2, #56] @ encoding: [0x9a,0xfd,0x1c,0x0f]
+vldrh.u32 q0, [r2, #56]
+
+# CHECK: vldrh.u32 q5, [r4, #56]! @ encoding: [0xbc,0xfd,0x1c,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r4, #56]! @ encoding: [0xbc,0xfd,0x1c,0xaf]
+vldrh.u32 q5, [r4, #56]!
+
+# CHECK: vldrh.u32 q5, [r4, #56]! @ encoding: [0xbc,0xfd,0x1c,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r4, #56]! @ encoding: [0xbc,0xfd,0x1c,0xaf]
+vldrh.u32 q5, [r4, #56]!
+
+# CHECK: vldrh.u32 q5, [r4], #-26 @ encoding: [0x3c,0xfc,0x0d,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r4], #-26 @ encoding: [0x3c,0xfc,0x0d,0xaf]
+vldrh.u32 q5, [r4], #-26
+
+# CHECK: vldrh.u32 q5, [r3], #-26 @ encoding: [0x3b,0xfc,0x0d,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r3], #-26 @ encoding: [0x3b,0xfc,0x0d,0xaf]
+vldrh.u32 q5, [r3], #-26
+
+# CHECK: vldrh.u32 q5, [r6, #-26] @ encoding: [0x1e,0xfd,0x0d,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r6, #-26] @ encoding: [0x1e,0xfd,0x0d,0xaf]
+vldrh.u32 q5, [r6, #-26]
+
+# CHECK: vldrh.u32 q5, [r6, #-64] @ encoding: [0x1e,0xfd,0x20,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r6, #-64] @ encoding: [0x1e,0xfd,0x20,0xaf]
+vldrh.u32 q5, [r6, #-64]
+
+# CHECK: vldrh.u32 q5, [r6, #-254] @ encoding: [0x1e,0xfd,0x7f,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r6, #-254] @ encoding: [0x1e,0xfd,0x7f,0xaf]
+vldrh.u32 q5, [r6, #-254]
+
+# CHECK: vldrh.u32 q5, [r4, #254]! @ encoding: [0xbc,0xfd,0x7f,0xaf]
+# CHECK-NOFP: vldrh.u32 q5, [r4, #254]! @ encoding: [0xbc,0xfd,0x7f,0xaf]
+vldrh.u32 q5, [r4, #254]!
+
+# CHECK: vldrh.s32 q0, [r0] @ encoding: [0x98,0xed,0x00,0x0f]
+# CHECK-NOFP: vldrh.s32 q0, [r0] @ encoding: [0x98,0xed,0x00,0x0f]
+vldrh.s32 q0, [r0]
+
+# CHECK: vldrh.s32 q1, [r0] @ encoding: [0x98,0xed,0x00,0x2f]
+# CHECK-NOFP: vldrh.s32 q1, [r0] @ encoding: [0x98,0xed,0x00,0x2f]
+vldrh.s32 q1, [r0]
+
+# CHECK: vldrh.s32 q0, [r7] @ encoding: [0x9f,0xed,0x00,0x0f]
+# CHECK-NOFP: vldrh.s32 q0, [r7] @ encoding: [0x9f,0xed,0x00,0x0f]
+vldrh.s32 q0, [r7]
+
+# CHECK: vldrh.s32 q3, [r7] @ encoding: [0x9f,0xed,0x00,0x6f]
+# CHECK-NOFP: vldrh.s32 q3, [r7] @ encoding: [0x9f,0xed,0x00,0x6f]
+vldrh.s32 q3, [r7]
+
+# CHECK: vldrh.s32 q0, [r4, #56] @ encoding: [0x9c,0xed,0x1c,0x0f]
+# CHECK-NOFP: vldrh.s32 q0, [r4, #56] @ encoding: [0x9c,0xed,0x1c,0x0f]
+vldrh.s32 q0, [r4, #56]
+
+# CHECK: vldrh.s32 q4, [r4, #56] @ encoding: [0x9c,0xed,0x1c,0x8f]
+# CHECK-NOFP: vldrh.s32 q4, [r4, #56] @ encoding: [0x9c,0xed,0x1c,0x8f]
+vldrh.s32 q4, [r4, #56]
+
+# CHECK: vldrh.s32 q0, [r2, #56] @ encoding: [0x9a,0xed,0x1c,0x0f]
+# CHECK-NOFP: vldrh.s32 q0, [r2, #56] @ encoding: [0x9a,0xed,0x1c,0x0f]
+vldrh.s32 q0, [r2, #56]
+
+# CHECK: vldrh.s32 q5, [r4, #56]! @ encoding: [0xbc,0xed,0x1c,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r4, #56]! @ encoding: [0xbc,0xed,0x1c,0xaf]
+vldrh.s32 q5, [r4, #56]!
+
+# CHECK: vldrh.s32 q5, [r4, #56]! @ encoding: [0xbc,0xed,0x1c,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r4, #56]! @ encoding: [0xbc,0xed,0x1c,0xaf]
+vldrh.s32 q5, [r4, #56]!
+
+# CHECK: vldrh.s32 q5, [r4], #-26 @ encoding: [0x3c,0xec,0x0d,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r4], #-26 @ encoding: [0x3c,0xec,0x0d,0xaf]
+vldrh.s32 q5, [r4], #-26
+
+# CHECK: vldrh.s32 q5, [r3], #-26 @ encoding: [0x3b,0xec,0x0d,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r3], #-26 @ encoding: [0x3b,0xec,0x0d,0xaf]
+vldrh.s32 q5, [r3], #-26
+
+# CHECK: vldrh.s32 q5, [r6, #-26] @ encoding: [0x1e,0xed,0x0d,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r6, #-26] @ encoding: [0x1e,0xed,0x0d,0xaf]
+vldrh.s32 q5, [r6, #-26]
+
+# CHECK: vldrh.s32 q5, [r6, #-64] @ encoding: [0x1e,0xed,0x20,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r6, #-64] @ encoding: [0x1e,0xed,0x20,0xaf]
+vldrh.s32 q5, [r6, #-64]
+
+# CHECK: vldrh.s32 q5, [r6, #-254] @ encoding: [0x1e,0xed,0x7f,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r6, #-254] @ encoding: [0x1e,0xed,0x7f,0xaf]
+vldrh.s32 q5, [r6, #-254]
+
+# CHECK: vldrh.s32 q5, [r4, #254]! @ encoding: [0xbc,0xed,0x7f,0xaf]
+# CHECK-NOFP: vldrh.s32 q5, [r4, #254]! @ encoding: [0xbc,0xed,0x7f,0xaf]
+vldrh.s32 q5, [r4, #254]!
+
+# CHECK: vstrh.32 q0, [r0] @ encoding: [0x88,0xed,0x00,0x0f]
+# CHECK-NOFP: vstrh.32 q0, [r0] @ encoding: [0x88,0xed,0x00,0x0f]
+vstrh.32 q0, [r0]
+
+# CHECK: vstrh.32 q1, [r0] @ encoding: [0x88,0xed,0x00,0x2f]
+# CHECK-NOFP: vstrh.32 q1, [r0] @ encoding: [0x88,0xed,0x00,0x2f]
+vstrh.32 q1, [r0]
+
+# CHECK: vstrh.32 q0, [r7] @ encoding: [0x8f,0xed,0x00,0x0f]
+# CHECK-NOFP: vstrh.32 q0, [r7] @ encoding: [0x8f,0xed,0x00,0x0f]
+vstrh.32 q0, [r7]
+
+# CHECK: vstrh.32 q3, [r7] @ encoding: [0x8f,0xed,0x00,0x6f]
+# CHECK-NOFP: vstrh.32 q3, [r7] @ encoding: [0x8f,0xed,0x00,0x6f]
+vstrh.32 q3, [r7]
+
+# CHECK: vstrh.32 q0, [r4, #56] @ encoding: [0x8c,0xed,0x1c,0x0f]
+# CHECK-NOFP: vstrh.32 q0, [r4, #56] @ encoding: [0x8c,0xed,0x1c,0x0f]
+vstrh.32 q0, [r4, #56]
+
+# CHECK: vstrh.32 q4, [r4, #56] @ encoding: [0x8c,0xed,0x1c,0x8f]
+# CHECK-NOFP: vstrh.32 q4, [r4, #56] @ encoding: [0x8c,0xed,0x1c,0x8f]
+vstrh.32 q4, [r4, #56]
+
+# CHECK: vstrh.32 q0, [r5, #56] @ encoding: [0x8d,0xed,0x1c,0x0f]
+# CHECK-NOFP: vstrh.32 q0, [r5, #56] @ encoding: [0x8d,0xed,0x1c,0x0f]
+vstrh.32 q0, [r5, #56]
+
+# CHECK: vstrh.32 q5, [r4, #56]! @ encoding: [0xac,0xed,0x1c,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r4, #56]! @ encoding: [0xac,0xed,0x1c,0xaf]
+vstrh.32 q5, [r4, #56]!
+
+# CHECK: vstrh.32 q5, [r4, #56]! @ encoding: [0xac,0xed,0x1c,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r4, #56]! @ encoding: [0xac,0xed,0x1c,0xaf]
+vstrh.32 q5, [r4, #56]!
+
+# CHECK: vstrh.32 q5, [r4], #-26 @ encoding: [0x2c,0xec,0x0d,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r4], #-26 @ encoding: [0x2c,0xec,0x0d,0xaf]
+vstrh.32 q5, [r4], #-26
+
+# CHECK: vstrh.32 q5, [r3], #-26 @ encoding: [0x2b,0xec,0x0d,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r3], #-26 @ encoding: [0x2b,0xec,0x0d,0xaf]
+vstrh.32 q5, [r3], #-26
+
+# CHECK: vstrh.32 q5, [r2, #-26] @ encoding: [0x0a,0xed,0x0d,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r2, #-26] @ encoding: [0x0a,0xed,0x0d,0xaf]
+vstrh.32 q5, [r2, #-26]
+
+# CHECK: vstrh.32 q5, [r2, #-64] @ encoding: [0x0a,0xed,0x20,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r2, #-64] @ encoding: [0x0a,0xed,0x20,0xaf]
+vstrh.32 q5, [r2, #-64]
+
+# CHECK: vstrh.32 q5, [r2, #-254] @ encoding: [0x0a,0xed,0x7f,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r2, #-254] @ encoding: [0x0a,0xed,0x7f,0xaf]
+vstrh.32 q5, [r2, #-254]
+
+# CHECK: vstrh.32 q5, [r4, #254]! @ encoding: [0xac,0xed,0x7f,0xaf]
+# CHECK-NOFP: vstrh.32 q5, [r4, #254]! @ encoding: [0xac,0xed,0x7f,0xaf]
+vstrh.32 q5, [r4, #254]!
+
+# CHECK: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+vldrw.u32 q0, [r0]
+
+# CHECK: vldrw.u32 q1, [r0] @ encoding: [0x90,0xed,0x00,0x3f]
+# CHECK-NOFP: vldrw.u32 q1, [r0] @ encoding: [0x90,0xed,0x00,0x3f]
+vldrw.u32 q1, [r0]
+
+# CHECK: vldrw.u32 q0, [r11] @ encoding: [0x9b,0xed,0x00,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r11] @ encoding: [0x9b,0xed,0x00,0x1f]
+vldrw.u32 q0, [r11]
+
+# CHECK: vldrw.u32 q3, [r11] @ encoding: [0x9b,0xed,0x00,0x7f]
+# CHECK-NOFP: vldrw.u32 q3, [r11] @ encoding: [0x9b,0xed,0x00,0x7f]
+vldrw.u32 q3, [r11]
+
+# CHECK: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+vldrw.u32 q0, [r4, #56]
+
+# CHECK: vldrw.u32 q4, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x9f]
+# CHECK-NOFP: vldrw.u32 q4, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x9f]
+vldrw.u32 q4, [r4, #56]
+
+# CHECK: vldrw.u32 q0, [r8, #56] @ encoding: [0x98,0xed,0x0e,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r8, #56] @ encoding: [0x98,0xed,0x0e,0x1f]
+vldrw.u32 q0, [r8, #56]
+
+# CHECK: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+vldrw.u32 q5, [r4, #56]!
+
+# CHECK: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+vldrw.u32 q5, [r4, #56]!
+
+# CHECK: vldrw.u32 q5, [r4], #-28 @ encoding: [0x34,0xec,0x07,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r4], #-28 @ encoding: [0x34,0xec,0x07,0xbf]
+vldrw.u32 q5, [r4], #-28
+
+# CHECK: vldrw.u32 q5, [r10], #-28 @ encoding: [0x3a,0xec,0x07,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r10], #-28 @ encoding: [0x3a,0xec,0x07,0xbf]
+vldrw.u32 q5, [r10], #-28
+
+# CHECK: vldrw.u32 q5, [sp, #-28] @ encoding: [0x1d,0xed,0x07,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [sp, #-28] @ encoding: [0x1d,0xed,0x07,0xbf]
+vldrw.u32 q5, [sp, #-28]
+
+# CHECK: vldrw.u32 q5, [sp, #-64] @ encoding: [0x1d,0xed,0x10,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [sp, #-64] @ encoding: [0x1d,0xed,0x10,0xbf]
+vldrw.u32 q5, [sp, #-64]
+
+# CHECK: vldrw.u32 q5, [sp, #-508] @ encoding: [0x1d,0xed,0x7f,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [sp, #-508] @ encoding: [0x1d,0xed,0x7f,0xbf]
+vldrw.u32 q5, [sp, #-508]
+
+# CHECK: vldrw.u32 q5, [r4, #508]! @ encoding: [0xb4,0xed,0x7f,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r4, #508]! @ encoding: [0xb4,0xed,0x7f,0xbf]
+vldrw.u32 q5, [r4, #508]!
+
+# CHECK: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+vstrw.32 q0, [r0]
+
+# CHECK: vstrw.32 q1, [r0] @ encoding: [0x80,0xed,0x00,0x3f]
+# CHECK-NOFP: vstrw.32 q1, [r0] @ encoding: [0x80,0xed,0x00,0x3f]
+vstrw.32 q1, [r0]
+
+# CHECK: vstrw.32 q0, [r11] @ encoding: [0x8b,0xed,0x00,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r11] @ encoding: [0x8b,0xed,0x00,0x1f]
+vstrw.32 q0, [r11]
+
+# CHECK: vstrw.32 q3, [r11] @ encoding: [0x8b,0xed,0x00,0x7f]
+# CHECK-NOFP: vstrw.32 q3, [r11] @ encoding: [0x8b,0xed,0x00,0x7f]
+vstrw.32 q3, [r11]
+
+# CHECK: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+vstrw.32 q0, [r4, #56]
+
+# CHECK: vstrw.32 q4, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x9f]
+# CHECK-NOFP: vstrw.32 q4, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x9f]
+vstrw.32 q4, [r4, #56]
+
+# CHECK: vstrw.32 q0, [r8, #56] @ encoding: [0x88,0xed,0x0e,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r8, #56] @ encoding: [0x88,0xed,0x0e,0x1f]
+vstrw.32 q0, [r8, #56]
+
+# CHECK: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+vstrw.32 q5, [r4, #56]!
+
+# CHECK: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+vstrw.32 q5, [r4, #56]!
+
+# CHECK: vstrw.32 q5, [r4], #-28 @ encoding: [0x24,0xec,0x07,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r4], #-28 @ encoding: [0x24,0xec,0x07,0xbf]
+vstrw.32 q5, [r4], #-28
+
+# CHECK: vstrw.32 q5, [r10], #-28 @ encoding: [0x2a,0xec,0x07,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r10], #-28 @ encoding: [0x2a,0xec,0x07,0xbf]
+vstrw.32 q5, [r10], #-28
+
+# CHECK: vstrw.32 q5, [sp, #-28] @ encoding: [0x0d,0xed,0x07,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [sp, #-28] @ encoding: [0x0d,0xed,0x07,0xbf]
+vstrw.32 q5, [sp, #-28]
+
+# CHECK: vstrw.32 q5, [sp, #-64] @ encoding: [0x0d,0xed,0x10,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [sp, #-64] @ encoding: [0x0d,0xed,0x10,0xbf]
+vstrw.32 q5, [sp, #-64]
+
+# CHECK: vstrw.32 q5, [sp, #-508] @ encoding: [0x0d,0xed,0x7f,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [sp, #-508] @ encoding: [0x0d,0xed,0x7f,0xbf]
+vstrw.32 q5, [sp, #-508]
+
+# CHECK: vstrw.32 q5, [r4, #508]! @ encoding: [0xa4,0xed,0x7f,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r4, #508]! @ encoding: [0xa4,0xed,0x7f,0xbf]
+vstrw.32 q5, [r4, #508]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrb.16 q0, [r8]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrh.u32 q0, [r8]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrw.32 q5, [sp, #-64]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrw.32 q5, [sp, #-3]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrw.32 q5, [sp, #512]
+
+# CHECK: vldrb.u8 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0e]
+# CHECK-NOFP: vldrb.u8 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0e]
+vldrb.u8 q0, [r0, q1]
+
+# CHECK: vldrb.u8 q3, [r10, q1] @ encoding: [0x9a,0xfc,0x02,0x6e]
+# CHECK-NOFP: vldrb.u8 q3, [r10, q1] @ encoding: [0x9a,0xfc,0x02,0x6e]
+vldrb.u8 q3, [r10, q1]
+
+# CHECK: vldrb.u16 q0, [r0, q1] @ encoding: [0x90,0xfc,0x82,0x0e]
+# CHECK-NOFP: vldrb.u16 q0, [r0, q1] @ encoding: [0x90,0xfc,0x82,0x0e]
+vldrb.u16 q0, [r0, q1]
+
+# CHECK: vldrb.u16 q3, [r9, q1] @ encoding: [0x99,0xfc,0x82,0x6e]
+# CHECK-NOFP: vldrb.u16 q3, [r9, q1] @ encoding: [0x99,0xfc,0x82,0x6e]
+vldrb.u16 q3, [r9, q1]
+
+# CHECK: vldrb.s16 q0, [r0, q1] @ encoding: [0x90,0xec,0x82,0x0e]
+# CHECK-NOFP: vldrb.s16 q0, [r0, q1] @ encoding: [0x90,0xec,0x82,0x0e]
+vldrb.s16 q0, [r0, q1]
+
+# CHECK: vldrb.s16 q3, [sp, q1] @ encoding: [0x9d,0xec,0x82,0x6e]
+# CHECK-NOFP: vldrb.s16 q3, [sp, q1] @ encoding: [0x9d,0xec,0x82,0x6e]
+vldrb.s16 q3, [sp, q1]
+
+# CHECK: vldrb.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0f]
+# CHECK-NOFP: vldrb.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0f]
+vldrb.u32 q0, [r0, q1]
+
+# CHECK: vldrb.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x6f]
+# CHECK-NOFP: vldrb.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x6f]
+vldrb.u32 q3, [r0, q1]
+
+# CHECK: vldrb.s32 q0, [r0, q1] @ encoding: [0x90,0xec,0x02,0x0f]
+# CHECK-NOFP: vldrb.s32 q0, [r0, q1] @ encoding: [0x90,0xec,0x02,0x0f]
+vldrb.s32 q0, [r0, q1]
+
+# CHECK: vldrb.s32 q3, [r0, q1] @ encoding: [0x90,0xec,0x02,0x6f]
+# CHECK-NOFP: vldrb.s32 q3, [r0, q1] @ encoding: [0x90,0xec,0x02,0x6f]
+vldrb.s32 q3, [r0, q1]
+
+# CHECK: vldrh.u16 q0, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x0e]
+# CHECK-NOFP: vldrh.u16 q0, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x0e]
+vldrh.u16 q0, [r0, q1]
+
+# CHECK: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+# CHECK-NOFP: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+vldrh.u16 q3, [r0, q1]
+
+# CHECK: vldrh.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x12,0x0f]
+# CHECK-NOFP: vldrh.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x12,0x0f]
+vldrh.u32 q0, [r0, q1]
+
+# CHECK: vldrh.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x12,0x6f]
+# CHECK-NOFP: vldrh.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x12,0x6f]
+vldrh.u32 q3, [r0, q1]
+
+# CHECK: vldrh.s32 q0, [r0, q1] @ encoding: [0x90,0xec,0x12,0x0f]
+# CHECK-NOFP: vldrh.s32 q0, [r0, q1] @ encoding: [0x90,0xec,0x12,0x0f]
+vldrh.s32 q0, [r0, q1]
+
+# CHECK: vldrh.s32 q3, [r0, q1] @ encoding: [0x90,0xec,0x12,0x6f]
+# CHECK-NOFP: vldrh.s32 q3, [r0, q1] @ encoding: [0x90,0xec,0x12,0x6f]
+vldrh.s32 q3, [r0, q1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrb.u8 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrb.u16 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrb.s16 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrb.u32 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrb.s32 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrh.u16 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrh.u32 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrh.s32 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrh.u16 q0, [r0, q0, uxtw #1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrh.u32 q0, [r0, q0, uxtw #1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrh.s32 q0, [r0, q0, uxtw #1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrw.u32 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrw.u32 q0, [r0, q0, uxtw #2]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrd.u64 q0, [r0, q0]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: destination vector register and vector offset register can't be identical
+vldrd.u64 q0, [r0, q0, uxtw #3]
+
+# CHECK: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+# CHECK-NOFP: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+vldrh.u16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+vldrw.u32 q0, [r0, q1]
+
+# CHECK: vldrw.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x6f]
+# CHECK-NOFP: vldrw.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x6f]
+vldrw.u32 q3, [r0, q1]
+
+# CHECK: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+vldrw.u32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vldrw.u32 q0, [sp, q1, uxtw #2] @ encoding: [0x9d,0xfc,0x43,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [sp, q1, uxtw #2] @ encoding: [0x9d,0xfc,0x43,0x0f]
+vldrw.u32 q0, [sp, q1, uxtw #2]
+
+# CHECK: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+vldrd.u64 q0, [r0, q1]
+
+# CHECK: vldrd.u64 q3, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x6f]
+# CHECK-NOFP: vldrd.u64 q3, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x6f]
+vldrd.u64 q3, [r0, q1]
+
+# CHECK: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+vldrd.u64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vldrd.u64 q0, [sp, q1, uxtw #3] @ encoding: [0x9d,0xfc,0xd3,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [sp, q1, uxtw #3] @ encoding: [0x9d,0xfc,0xd3,0x0f]
+vldrd.u64 q0, [sp, q1, uxtw #3]
+
+# CHECK: vstrb.8 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0e]
+# CHECK-NOFP: vstrb.8 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0e]
+vstrb.8 q0, [r0, q1]
+
+# CHECK: vstrb.8 q3, [r10, q1] @ encoding: [0x8a,0xec,0x02,0x6e]
+# CHECK-NOFP: vstrb.8 q3, [r10, q1] @ encoding: [0x8a,0xec,0x02,0x6e]
+vstrb.8 q3, [r10, q1]
+
+# CHECK: vstrb.8 q3, [r0, q3] @ encoding: [0x80,0xec,0x06,0x6e]
+# CHECK-NOFP: vstrb.8 q3, [r0, q3] @ encoding: [0x80,0xec,0x06,0x6e]
+vstrb.8 q3, [r0, q3]
+
+# CHECK: vstrb.16 q0, [r0, q1] @ encoding: [0x80,0xec,0x82,0x0e]
+# CHECK-NOFP: vstrb.16 q0, [r0, q1] @ encoding: [0x80,0xec,0x82,0x0e]
+vstrb.16 q0, [r0, q1]
+
+# CHECK: vstrb.16 q3, [sp, q1] @ encoding: [0x8d,0xec,0x82,0x6e]
+# CHECK-NOFP: vstrb.16 q3, [sp, q1] @ encoding: [0x8d,0xec,0x82,0x6e]
+vstrb.16 q3, [sp, q1]
+
+# CHECK: vstrb.16 q3, [r0, q3] @ encoding: [0x80,0xec,0x86,0x6e]
+# CHECK-NOFP: vstrb.16 q3, [r0, q3] @ encoding: [0x80,0xec,0x86,0x6e]
+vstrb.16 q3, [r0, q3]
+
+# CHECK: vstrb.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0f]
+# CHECK-NOFP: vstrb.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0f]
+vstrb.32 q0, [r0, q1]
+
+# CHECK: vstrb.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x02,0x6f]
+# CHECK-NOFP: vstrb.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x02,0x6f]
+vstrb.32 q3, [r0, q1]
+
+# CHECK: vstrb.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x06,0x6f]
+# CHECK-NOFP: vstrb.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x06,0x6f]
+vstrb.32 q3, [r0, q3]
+
+# CHECK: vstrh.16 q0, [r0, q1] @ encoding: [0x80,0xec,0x92,0x0e]
+# CHECK-NOFP: vstrh.16 q0, [r0, q1] @ encoding: [0x80,0xec,0x92,0x0e]
+vstrh.16 q0, [r0, q1]
+
+# CHECK: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+# CHECK-NOFP: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+vstrh.16 q3, [r0, q1]
+
+# CHECK: vstrh.16 q3, [r0, q3] @ encoding: [0x80,0xec,0x96,0x6e]
+# CHECK-NOFP: vstrh.16 q3, [r0, q3] @ encoding: [0x80,0xec,0x96,0x6e]
+vstrh.16 q3, [r0, q3]
+
+# CHECK: vstrh.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x12,0x0f]
+# CHECK-NOFP: vstrh.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x12,0x0f]
+vstrh.32 q0, [r0, q1]
+
+# CHECK: vstrh.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x12,0x6f]
+# CHECK-NOFP: vstrh.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x12,0x6f]
+vstrh.32 q3, [r0, q1]
+
+# CHECK: vstrh.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x16,0x6f]
+# CHECK-NOFP: vstrh.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x16,0x6f]
+vstrh.32 q3, [r0, q3]
+
+# CHECK: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+# CHECK-NOFP: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+vstrh.16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vstrh.32 q3, [r8, q3, uxtw #1] @ encoding: [0x88,0xec,0x17,0x6f]
+# CHECK-NOFP: vstrh.32 q3, [r8, q3, uxtw #1] @ encoding: [0x88,0xec,0x17,0x6f]
+vstrh.32 q3, [r8, q3, uxtw #1]
+
+# CHECK: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+vstrw.32 q0, [r0, q1]
+
+# CHECK: vstrw.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x42,0x6f]
+# CHECK-NOFP: vstrw.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x42,0x6f]
+vstrw.32 q3, [r0, q1]
+
+# CHECK: vstrw.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x46,0x6f]
+# CHECK-NOFP: vstrw.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x46,0x6f]
+vstrw.32 q3, [r0, q3]
+
+# CHECK: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+vstrw.32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vstrw.32 q0, [sp, q1, uxtw #2] @ encoding: [0x8d,0xec,0x43,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [sp, q1, uxtw #2] @ encoding: [0x8d,0xec,0x43,0x0f]
+vstrw.32 q0, [sp, q1, uxtw #2]
+
+# CHECK: vstrd.64 q0, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x0f]
+# CHECK-NOFP: vstrd.64 q0, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x0f]
+vstrd.64 q0, [r0, q1]
+
+# CHECK: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+# CHECK-NOFP: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+vstrd.64 q3, [r0, q1]
+
+# CHECK: vstrd.64 q3, [r0, q3] @ encoding: [0x80,0xec,0xd6,0x6f]
+# CHECK-NOFP: vstrd.64 q3, [r0, q3] @ encoding: [0x80,0xec,0xd6,0x6f]
+vstrd.64 q3, [r0, q3]
+
+# CHECK: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+# CHECK-NOFP: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+vstrd.64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vstrd.64 q0, [sp, q1, uxtw #3] @ encoding: [0x8d,0xec,0xd3,0x0f]
+# CHECK-NOFP: vstrd.64 q0, [sp, q1, uxtw #3] @ encoding: [0x8d,0xec,0xd3,0x0f]
+vstrd.64 q0, [sp, q1, uxtw #3]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: operand must be a register in range [q0, q7]
+vstrw.32 q9, [sp, q1, uxtw #2]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrh.16 q3, [pc, q1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrd.64 q0, [r0, q1, uxtw #1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrd.u64 q0, [r0, q1, uxtw #1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrd.64 q0, [r0, q1, uxtw #2]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrd.u64 q0, [r0, q1, uxtw #2]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrw.u32 q0, [r0, q1, uxtw #1]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrh.16 q0, [r0, q1, uxtw #2]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrb.32 q0, [r11, q1, uxtw #1]
+
+# CHECK: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+# CHECK-NOFP: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+vldrw.u32 q0, [q1]
+
+# CHECK: vldrw.u32 q7, [q1] @ encoding: [0x92,0xfd,0x00,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1] @ encoding: [0x92,0xfd,0x00,0xfe]
+vldrw.u32 q7, [q1]
+
+# CHECK: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+vldrw.u32 q7, [q1]!
+
+# CHECK: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+vldrw.u32 q7, [q1, #4]
+
+# CHECK: vldrw.u32 q7, [q1, #-4] @ encoding: [0x12,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #-4] @ encoding: [0x12,0xfd,0x01,0xfe]
+vldrw.u32 q7, [q1, #-4]
+
+# CHECK: vldrw.u32 q7, [q1, #508] @ encoding: [0x92,0xfd,0x7f,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #508] @ encoding: [0x92,0xfd,0x7f,0xfe]
+vldrw.u32 q7, [q1, #508]
+
+# CHECK: vldrw.u32 q7, [q1, #-508] @ encoding: [0x12,0xfd,0x7f,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #-508] @ encoding: [0x12,0xfd,0x7f,0xfe]
+vldrw.u32 q7, [q1, #-508]
+
+# CHECK: vldrw.u32 q7, [q1, #264] @ encoding: [0x92,0xfd,0x42,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #264] @ encoding: [0x92,0xfd,0x42,0xfe]
+vldrw.u32 q7, [q1, #264]
+
+# CHECK: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+vldrw.u32 q7, [q1, #4]!
+
+# CHECK: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+# CHECK-NOFP: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+vstrw.32 q0, [q1]
+
+# CHECK: vstrw.32 q7, [q1] @ encoding: [0x82,0xfd,0x00,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1] @ encoding: [0x82,0xfd,0x00,0xfe]
+vstrw.32 q7, [q1]
+
+# CHECK: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+vstrw.32 q7, [q1]!
+
+# CHECK: vstrw.32 q7, [q7] @ encoding: [0x8e,0xfd,0x00,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q7] @ encoding: [0x8e,0xfd,0x00,0xfe]
+vstrw.32 q7, [q7]
+
+# CHECK: vstrw.32 q7, [q1, #4] @ encoding: [0x82,0xfd,0x01,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #4] @ encoding: [0x82,0xfd,0x01,0xfe]
+vstrw.32 q7, [q1, #4]
+
+# CHECK: vstrw.32 q7, [q1, #-4] @ encoding: [0x02,0xfd,0x01,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #-4] @ encoding: [0x02,0xfd,0x01,0xfe]
+vstrw.32 q7, [q1, #-4]
+
+# CHECK: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+vstrw.32 q7, [q1, #508]
+
+# CHECK: vstrw.32 q7, [q1, #-508] @ encoding: [0x02,0xfd,0x7f,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #-508] @ encoding: [0x02,0xfd,0x7f,0xfe]
+vstrw.32 q7, [q1, #-508]
+
+# CHECK: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+vstrw.32 q7, [q1, #264]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: operand must be a register in range [q0, q7]
+vstrw.32 q8, [q1]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrw.32 q4, [q1, #3]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vldrw.u32 q7, [q1, #512]
+
+# CHECK: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+# CHECK-NOFP: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+vldrd.u64 q0, [q1]
+
+# CHECK: vldrd.u64 q7, [q1] @ encoding: [0x92,0xfd,0x00,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1] @ encoding: [0x92,0xfd,0x00,0xff]
+vldrd.u64 q7, [q1]
+
+# CHECK: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+vldrd.u64 q7, [q1]!
+
+# CHECK: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+vldrd.u64 q7, [q1, #8]
+
+# CHECK: vldrd.u64 q7, [q1, #-8] @ encoding: [0x12,0xfd,0x01,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #-8] @ encoding: [0x12,0xfd,0x01,0xff]
+vldrd.u64 q7, [q1, #-8]
+
+# CHECK: vldrd.u64 q7, [q1, #1016] @ encoding: [0x92,0xfd,0x7f,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #1016] @ encoding: [0x92,0xfd,0x7f,0xff]
+vldrd.u64 q7, [q1, #1016]
+
+# CHECK: vldrd.u64 q7, [q1, #-1016] @ encoding: [0x12,0xfd,0x7f,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #-1016] @ encoding: [0x12,0xfd,0x7f,0xff]
+vldrd.u64 q7, [q1, #-1016]
+
+# CHECK: vldrd.u64 q7, [q1, #264] @ encoding: [0x92,0xfd,0x21,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #264] @ encoding: [0x92,0xfd,0x21,0xff]
+vldrd.u64 q7, [q1, #264]
+
+# CHECK: vldrd.u64 q7, [q1, #624] @ encoding: [0x92,0xfd,0x4e,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #624] @ encoding: [0x92,0xfd,0x4e,0xff]
+vldrd.u64 q7, [q1, #624]
+
+# CHECK: vldrd.u64 q7, [q1, #264] @ encoding: [0x92,0xfd,0x21,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #264] @ encoding: [0x92,0xfd,0x21,0xff]
+vldrd.u64 q7, [q1, #264]
+
+# CHECK: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+vldrd.u64 q7, [q1, #-1016]!
+
+# CHECK: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+# CHECK-NOFP: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+vstrd.64 q0, [q1]
+
+# CHECK: vstrd.64 q7, [q1] @ encoding: [0x82,0xfd,0x00,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1] @ encoding: [0x82,0xfd,0x00,0xff]
+vstrd.64 q7, [q1]
+
+# CHECK: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+vstrd.64 q7, [q1]!
+
+# CHECK: vstrd.64 q7, [q7] @ encoding: [0x8e,0xfd,0x00,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q7] @ encoding: [0x8e,0xfd,0x00,0xff]
+vstrd.64 q7, [q7]
+
+# CHECK: vstrd.64 q7, [q1, #8] @ encoding: [0x82,0xfd,0x01,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #8] @ encoding: [0x82,0xfd,0x01,0xff]
+vstrd.64 q7, [q1, #8]
+
+# CHECK: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+vstrd.64 q7, [q1, #-8]!
+
+# CHECK: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+vstrd.64 q7, [q1, #1016]
+
+# CHECK: vstrd.64 q7, [q1, #-1016] @ encoding: [0x02,0xfd,0x7f,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #-1016] @ encoding: [0x02,0xfd,0x7f,0xff]
+vstrd.64 q7, [q1, #-1016]
+
+# CHECK: vstrd.64 q7, [q1, #264] @ encoding: [0x82,0xfd,0x21,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #264] @ encoding: [0x82,0xfd,0x21,0xff]
+vstrd.64 q7, [q1, #264]
+
+# CHECK: vstrd.64 q7, [q1, #624] @ encoding: [0x82,0xfd,0x4e,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #624] @ encoding: [0x82,0xfd,0x4e,0xff]
+vstrd.64 q7, [q1, #624]
+
+# CHECK: vstrd.64 q7, [q1, #264] @ encoding: [0x82,0xfd,0x21,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #264] @ encoding: [0x82,0xfd,0x21,0xff]
+vstrd.64 q7, [q1, #264]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: operand must be a register in range [q0, q7]
+vldrd.u64 q8, [q1]!
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrd.64 q7, [q1, #1024]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrd.64 q4, [q1, #3]
+
+# ERROR: [[@LINE+1]]:{{[0-9]+}}: {{error|note}}: invalid operand for instruction
+vstrd.64 q4, [q1, #4]
+
+# CHECK: vldrb.u8 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1e]
+vldrb.s8 q0, [r0]
+
+# CHECK: vldrb.u8 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1e]
+vldrb.8 q0, [r0]
+
+# CHECK: vldrb.u8 q0, [r8, #56] @ encoding: [0x98,0xed,0x38,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r8, #56] @ encoding: [0x98,0xed,0x38,0x1e]
+vldrb.s8 q0, [r8, #56]
+
+# CHECK: vldrb.u8 q0, [r8, #56] @ encoding: [0x98,0xed,0x38,0x1e]
+# CHECK-NOFP: vldrb.u8 q0, [r8, #56] @ encoding: [0x98,0xed,0x38,0x1e]
+vldrb.8 q0, [r8, #56]
+
+# CHECK: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+vldrb.s8 q5, [r4, #56]!
+
+# CHECK: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+# CHECK-NOFP: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+vldrb.8 q5, [r4, #56]!
+
+# CHECK: vstrb.8 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1e]
+# CHECK-NOFP: vstrb.8 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1e]
+vstrb.u8 q0, [r0]
+
+# CHECK: vstrb.8 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1e]
+# CHECK-NOFP: vstrb.8 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1e]
+vstrb.s8 q0, [r0]
+
+# CHECK: vstrb.8 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x9e]
+# CHECK-NOFP: vstrb.8 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x9e]
+vstrb.u8 q4, [r4, #56]
+
+# CHECK: vstrb.8 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x9e]
+# CHECK-NOFP: vstrb.8 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x9e]
+vstrb.s8 q4, [r4, #56]
+
+# CHECK: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+vstrb.u8 q5, [r4, #56]!
+
+# CHECK: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+# CHECK-NOFP: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+vstrb.s8 q5, [r4, #56]!
+
+# CHECK: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+vldrh.s16 q0, [r0]
+
+# CHECK: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+vldrh.f16 q0, [r0]
+
+# CHECK: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+vldrh.16 q0, [r0]
+
+# CHECK: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+vldrh.s16 q0, [r4, #56]
+
+# CHECK: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+vldrh.f16 q0, [r4, #56]
+
+# CHECK: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+# CHECK-NOFP: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+vldrh.16 q0, [r4, #56]
+
+# CHECK: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+vldrh.s16 q5, [r4, #56]!
+
+# CHECK: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+vldrh.f16 q5, [r4, #56]!
+
+# CHECK: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+vldrh.16 q5, [r4, #56]!
+
+# CHECK: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+vstrh.u16 q0, [r0]
+
+# CHECK: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+vstrh.s16 q0, [r0]
+
+# CHECK: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+vstrh.f16 q0, [r0]
+
+# CHECK: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+vstrh.u16 q0, [r4, #56]
+
+# CHECK: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+vstrh.s16 q0, [r4, #56]
+
+# CHECK: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+# CHECK-NOFP: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+vstrh.f16 q0, [r4, #56]
+
+# CHECK: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+vstrh.u16 q5, [r4, #56]!
+
+# CHECK: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+vstrh.s16 q5, [r4, #56]!
+
+# CHECK: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+# CHECK-NOFP: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+vstrh.f16 q5, [r4, #56]!
+
+# CHECK: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+vldrw.s32 q0, [r0]
+
+# CHECK: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+vldrw.f32 q0, [r0]
+
+# CHECK: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+vldrw.32 q0, [r0]
+
+# CHECK: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+vldrw.s32 q0, [r4, #56]
+
+# CHECK: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+vldrw.f32 q0, [r4, #56]
+
+# CHECK: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+# CHECK-NOFP: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+vldrw.32 q0, [r4, #56]
+
+# CHECK: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+vldrw.s32 q5, [r4, #56]!
+
+# CHECK: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+vldrw.f32 q5, [r4, #56]!
+
+# CHECK: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+vldrw.32 q5, [r4, #56]!
+
+# CHECK: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+vstrw.u32 q0, [r0]
+
+# CHECK: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+vstrw.s32 q0, [r0]
+
+# CHECK: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+vstrw.f32 q0, [r0]
+
+# CHECK: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+vstrw.u32 q0, [r4, #56]
+
+# CHECK: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+vstrw.s32 q0, [r4, #56]
+
+# CHECK: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+# CHECK-NOFP: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+vstrw.f32 q0, [r4, #56]
+
+# CHECK: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+vstrw.u32 q5, [r4, #56]!
+
+# CHECK: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+vstrw.s32 q5, [r4, #56]!
+
+# CHECK: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+# CHECK-NOFP: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+vstrw.f32 q5, [r4, #56]!
+
+# CHECK: vldrb.u8 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0e]
+# CHECK-NOFP: vldrb.u8 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0e]
+vldrb.s8 q0, [r0, q1]
+
+# CHECK: vldrb.u8 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0e]
+# CHECK-NOFP: vldrb.u8 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0e]
+vldrb.8 q0, [r0, q1]
+
+# CHECK: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+# CHECK-NOFP: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+vldrh.s16 q3, [r0, q1]
+
+# CHECK: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+# CHECK-NOFP: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+vldrh.f16 q3, [r0, q1]
+
+# CHECK: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+# CHECK-NOFP: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+vldrh.16 q3, [r0, q1]
+
+# CHECK: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+# CHECK-NOFP: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+vldrh.s16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+# CHECK-NOFP: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+vldrh.f16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+# CHECK-NOFP: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+vldrh.16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+vldrw.s32 q0, [r0, q1]
+
+# CHECK: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+vldrw.f32 q0, [r0, q1]
+
+# CHECK: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+vldrw.32 q0, [r0, q1]
+
+# CHECK: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+vldrw.s32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+vldrw.f32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+# CHECK-NOFP: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+vldrw.32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+vldrd.s64 q0, [r0, q1]
+
+# CHECK: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+vldrd.f64 q0, [r0, q1]
+
+# CHECK: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+vldrd.64 q0, [r0, q1]
+
+# CHECK: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+vldrd.s64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+vldrd.f64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+# CHECK-NOFP: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+vldrd.64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vstrb.8 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0e]
+# CHECK-NOFP: vstrb.8 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0e]
+vstrb.u8 q0, [r0, q1]
+
+# CHECK: vstrb.8 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0e]
+# CHECK-NOFP: vstrb.8 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0e]
+vstrb.s8 q0, [r0, q1]
+
+# CHECK: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+# CHECK-NOFP: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+vstrh.u16 q3, [r0, q1]
+
+# CHECK: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+# CHECK-NOFP: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+vstrh.s16 q3, [r0, q1]
+
+# CHECK: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+# CHECK-NOFP: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+vstrh.f16 q3, [r0, q1]
+
+# CHECK: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+# CHECK-NOFP: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+vstrh.u16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+# CHECK-NOFP: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+vstrh.s16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+# CHECK-NOFP: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+vstrh.f16 q0, [r0, q1, uxtw #1]
+
+# CHECK: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+vstrw.u32 q0, [r0, q1]
+
+# CHECK: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+vstrw.s32 q0, [r0, q1]
+
+# CHECK: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+vstrw.f32 q0, [r0, q1]
+
+# CHECK: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+vstrw.u32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+vstrw.s32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+# CHECK-NOFP: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+vstrw.f32 q0, [r0, q1, uxtw #2]
+
+# CHECK: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+# CHECK-NOFP: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+vstrd.u64 q3, [r0, q1]
+
+# CHECK: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+# CHECK-NOFP: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+vstrd.s64 q3, [r0, q1]
+
+# CHECK: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+# CHECK-NOFP: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+vstrd.f64 q3, [r0, q1]
+
+# CHECK: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+# CHECK-NOFP: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+vstrd.u64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+# CHECK-NOFP: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+vstrd.s64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+# CHECK-NOFP: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+vstrd.f64 q0, [r0, q1, uxtw #3]
+
+# CHECK: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+# CHECK-NOFP: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+vldrw.s32 q0, [q1]
+
+# CHECK: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+# CHECK-NOFP: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+vldrw.f32 q0, [q1]
+
+# CHECK: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+# CHECK-NOFP: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+vldrw.32 q0, [q1]
+
+# CHECK: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+vldrw.s32 q7, [q1]!
+
+# CHECK: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+vldrw.f32 q7, [q1]!
+
+# CHECK: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+vldrw.32 q7, [q1]!
+
+# CHECK: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+vldrw.s32 q7, [q1, #4]
+
+# CHECK: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+vldrw.f32 q7, [q1, #4]
+
+# CHECK: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+vldrw.32 q7, [q1, #4]
+
+# CHECK: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+vldrw.s32 q7, [q1, #4]!
+
+# CHECK: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+vldrw.f32 q7, [q1, #4]!
+
+# CHECK: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+# CHECK-NOFP: vldrw.u32 q7, [q1, #4]! @ encoding: [0xb2,0xfd,0x01,0xfe]
+vldrw.u32 q7, [q1, #4]!
+
+# CHECK: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+# CHECK-NOFP: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+vstrw.u32 q0, [q1]
+
+# CHECK: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+# CHECK-NOFP: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+vstrw.s32 q0, [q1]
+
+# CHECK: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+# CHECK-NOFP: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+vstrw.f32 q0, [q1]
+
+# CHECK: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+vstrw.u32 q7, [q1]!
+
+# CHECK: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+vstrw.s32 q7, [q1]!
+
+# CHECK: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+vstrw.f32 q7, [q1]!
+
+# CHECK: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+vstrw.u32 q7, [q1, #508]
+
+# CHECK: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+vstrw.s32 q7, [q1, #508]
+
+# CHECK: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+vstrw.f32 q7, [q1, #508]
+
+# CHECK: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+vstrw.u32 q7, [q1, #264]!
+
+# CHECK: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+vstrw.s32 q7, [q1, #264]!
+
+# CHECK: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+# CHECK-NOFP: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+vstrw.f32 q7, [q1, #264]!
+
+# CHECK: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+# CHECK-NOFP: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+vldrd.s64 q0, [q1]
+
+# CHECK: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+# CHECK-NOFP: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+vldrd.f64 q0, [q1]
+
+# CHECK: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+# CHECK-NOFP: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+vldrd.64 q0, [q1]
+
+# CHECK: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+vldrd.s64 q7, [q1]!
+
+# CHECK: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+vldrd.f64 q7, [q1]!
+
+# CHECK: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+vldrd.64 q7, [q1]!
+
+# CHECK: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+vldrd.s64 q7, [q1, #8]
+
+# CHECK: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+vldrd.f64 q7, [q1, #8]
+
+# CHECK: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+vldrd.64 q7, [q1, #8]
+
+# CHECK: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+vldrd.s64 q7, [q1, #-1016]!
+
+# CHECK: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+vldrd.f64 q7, [q1, #-1016]!
+
+# CHECK: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+# CHECK-NOFP: vldrd.u64 q7, [q1, #-1016]! @ encoding: [0x32,0xfd,0x7f,0xff]
+vldrd.64 q7, [q1, #-1016]!
+
+# CHECK: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+# CHECK-NOFP: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+vstrd.u64 q0, [q1]
+
+# CHECK: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+# CHECK-NOFP: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+vstrd.s64 q0, [q1]
+
+# CHECK: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+# CHECK-NOFP: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+vstrd.f64 q0, [q1]
+
+# CHECK: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+vstrd.u64 q7, [q1]!
+
+# CHECK: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+vstrd.s64 q7, [q1]!
+
+# CHECK: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+vstrd.f64 q7, [q1]!
+
+# CHECK: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+vstrd.u64 q7, [q1, #1016]
+
+# CHECK: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+vstrd.s64 q7, [q1, #1016]
+
+# CHECK: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+vstrd.f64 q7, [q1, #1016]
+
+# CHECK: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+vstrd.u64 q7, [q1, #-8]!
+
+# CHECK: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+vstrd.s64 q7, [q1, #-8]!
+
+# CHECK: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+# CHECK-NOFP: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+vstrd.f64 q7, [q1, #-8]!
+
+vpste
+vstrwt.f32 q7, [q1, #264]!
+vldrde.64 q7, [q1, #8]
+# CHECK: vpste @ encoding: [0x71,0xfe,0x4d,0x8f]
+# CHECK-NOFP: vpste @ encoding: [0x71,0xfe,0x4d,0x8f]
+# CHECK: vstrwt.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+# CHECK-NOFP: vstrwt.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+# CHECK: vldrde.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+# CHECK-NOFP: vldrde.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]

llvm/test/MC/Disassembler/ARM/mve-load-store.txt

+# RUN: llvm-mc -disassemble -triple=thumbv8.1m.main-none-eabi -mattr=+mve.fp,+fp64 -show-encoding %s 2> %t | FileCheck %s
+# RUN: FileCheck --check-prefix=ERROR < %t %s
+# RUN: not llvm-mc -disassemble -triple=thumbv8.1m.main-none-eabi -show-encoding %s &> %t
+# RUN: FileCheck --check-prefix=CHECK-NOMVE < %t %s
+
+# CHECK: vldrb.u8 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x00,0x1e]
+
+# CHECK: vldrb.u8 q1, [r0] @ encoding: [0x90,0xed,0x00,0x3e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x00,0x3e]
+
+# CHECK: vldrb.u8 q0, [r11] @ encoding: [0x9b,0xed,0x00,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9b,0xed,0x00,0x1e]
+
+# CHECK: vldrb.u8 q3, [r11] @ encoding: [0x9b,0xed,0x00,0x7e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9b,0xed,0x00,0x7e]
+
+# CHECK: vldrb.u8 q0, [r4, #56] @ encoding: [0x94,0xed,0x38,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x38,0x1e]
+
+# CHECK: vldrb.u8 q4, [r4, #56] @ encoding: [0x94,0xed,0x38,0x9e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x38,0x9e]
+
+# CHECK: vldrb.u8 q0, [r8, #56] @ encoding: [0x98,0xed,0x38,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xed,0x38,0x1e]
+
+# CHECK: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x38,0xbe]
+
+# CHECK: vldrb.u8 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x38,0xbe]
+
+# CHECK: vldrb.u8 q5, [r4], #-25 @ encoding: [0x34,0xec,0x19,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x34,0xec,0x19,0xbe]
+
+# CHECK: vldrb.u8 q5, [r10], #-25 @ encoding: [0x3a,0xec,0x19,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x3a,0xec,0x19,0xbe]
+
+# CHECK: vldrb.u8 q5, [sp, #-25] @ encoding: [0x1d,0xed,0x19,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0x19,0xbe]
+
+# CHECK: vldrb.u8 q5, [sp, #-64] @ encoding: [0x1d,0xed,0x40,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0x40,0xbe]
+
+# CHECK: vstrb.8 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x00,0x1e]
+
+# CHECK: vstrb.8 q1, [r0] @ encoding: [0x80,0xed,0x00,0x3e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x00,0x3e]
+
+# CHECK: vstrb.8 q0, [r11] @ encoding: [0x8b,0xed,0x00,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8b,0xed,0x00,0x1e]
+
+# CHECK: vstrb.8 q3, [r11] @ encoding: [0x8b,0xed,0x00,0x7e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8b,0xed,0x00,0x7e]
+
+# CHECK: vstrb.8 q0, [r4, #56] @ encoding: [0x84,0xed,0x38,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x38,0x1e]
+
+# CHECK: vstrb.8 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x9e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x38,0x9e]
+
+# CHECK: vstrb.8 q0, [r8, #56] @ encoding: [0x88,0xed,0x38,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x88,0xed,0x38,0x1e]
+
+# CHECK: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x38,0xbe]
+
+# CHECK: vstrb.8 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x38,0xbe]
+
+# CHECK: vstrb.8 q5, [r4], #-25 @ encoding: [0x24,0xec,0x19,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x24,0xec,0x19,0xbe]
+
+# CHECK: vstrb.8 q5, [r10], #-25 @ encoding: [0x2a,0xec,0x19,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x2a,0xec,0x19,0xbe]
+
+# CHECK: vstrb.8 q5, [sp, #-25] @ encoding: [0x0d,0xed,0x19,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0x19,0xbe]
+
+# CHECK: vstrb.8 q5, [sp, #-64] @ encoding: [0x0d,0xed,0x40,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0x40,0xbe]
+
+# ERROR: [[@LINE+1]]:2: warning: potentially undefined instruction encoding
+[0xad,0xec,0x00,0x1e]
+
+# CHECK: vldrb.u16 q0, [r0] @ encoding: [0x90,0xfd,0x80,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfd,0x80,0x0e]
+
+# CHECK: vldrb.u16 q1, [r0] @ encoding: [0x90,0xfd,0x80,0x2e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfd,0x80,0x2e]
+
+# CHECK: vldrb.u16 q0, [r7] @ encoding: [0x97,0xfd,0x80,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xfd,0x80,0x0e]
+
+# CHECK: vldrb.u16 q3, [r7] @ encoding: [0x97,0xfd,0x80,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xfd,0x80,0x6e]
+
+# CHECK: vldrb.u16 q0, [r4, #56] @ encoding: [0x94,0xfd,0xb8,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xfd,0xb8,0x0e]
+
+# CHECK: vldrb.u16 q4, [r4, #56] @ encoding: [0x94,0xfd,0xb8,0x8e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xfd,0xb8,0x8e]
+
+# CHECK: vldrb.u16 q0, [r2, #56] @ encoding: [0x92,0xfd,0xb8,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0xb8,0x0e]
+
+# CHECK: vldrb.u16 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0xb8,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xfd,0xb8,0xae]
+
+# CHECK: vldrb.u16 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0xb8,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xfd,0xb8,0xae]
+
+# CHECK: vldrb.u16 q5, [r4], #-25 @ encoding: [0x34,0xfc,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x34,0xfc,0x99,0xae]
+
+# CHECK: vldrb.u16 q5, [r3], #-25 @ encoding: [0x33,0xfc,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x33,0xfc,0x99,0xae]
+
+# CHECK: vldrb.u16 q5, [r6, #-25] @ encoding: [0x16,0xfd,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xfd,0x99,0xae]
+
+# CHECK: vldrb.u16 q5, [r6, #-64] @ encoding: [0x16,0xfd,0xc0,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xfd,0xc0,0xae]
+
+# CHECK: vldrb.s16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x80,0x0e]
+
+# CHECK: vldrb.s16 q1, [r0] @ encoding: [0x90,0xed,0x80,0x2e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x80,0x2e]
+
+# CHECK: vldrb.s16 q0, [r7] @ encoding: [0x97,0xed,0x80,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xed,0x80,0x0e]
+
+# CHECK: vldrb.s16 q3, [r7] @ encoding: [0x97,0xed,0x80,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xed,0x80,0x6e]
+
+# CHECK: vldrb.s16 q0, [r4, #56] @ encoding: [0x94,0xed,0xb8,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0xb8,0x0e]
+
+# CHECK: vldrb.s16 q4, [r4, #56] @ encoding: [0x94,0xed,0xb8,0x8e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0xb8,0x8e]
+
+# CHECK: vldrb.s16 q0, [r2, #56] @ encoding: [0x92,0xed,0xb8,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xed,0xb8,0x0e]
+
+# CHECK: vldrb.s16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0xb8,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0xb8,0xae]
+
+# CHECK: vldrb.s16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0xb8,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0xb8,0xae]
+
+# CHECK: vldrb.s16 q5, [r4], #-25 @ encoding: [0x34,0xec,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x34,0xec,0x99,0xae]
+
+# CHECK: vldrb.s16 q5, [r3], #-25 @ encoding: [0x33,0xec,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x33,0xec,0x99,0xae]
+
+# CHECK: vldrb.s16 q5, [r6, #-25] @ encoding: [0x16,0xed,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xed,0x99,0xae]
+
+# CHECK: vldrb.s16 q5, [r6, #-64] @ encoding: [0x16,0xed,0xc0,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xed,0xc0,0xae]
+
+# CHECK: vstrb.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x80,0x0e]
+
+# CHECK: vstrb.16 q1, [r0] @ encoding: [0x80,0xed,0x80,0x2e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x80,0x2e]
+
+# CHECK: vstrb.16 q0, [r7] @ encoding: [0x87,0xed,0x80,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x87,0xed,0x80,0x0e]
+
+# CHECK: vstrb.16 q3, [r7] @ encoding: [0x87,0xed,0x80,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x87,0xed,0x80,0x6e]
+
+# CHECK: vstrb.16 q0, [r4, #56] @ encoding: [0x84,0xed,0xb8,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0xb8,0x0e]
+
+# CHECK: vstrb.16 q4, [r4, #56] @ encoding: [0x84,0xed,0xb8,0x8e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0xb8,0x8e]
+
+# CHECK: vstrb.16 q0, [r5, #56] @ encoding: [0x85,0xed,0xb8,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x85,0xed,0xb8,0x0e]
+
+# CHECK: vstrb.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0xb8,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0xb8,0xae]
+
+# CHECK: vstrb.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0xb8,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0xb8,0xae]
+
+# CHECK: vstrb.16 q5, [r4], #-25 @ encoding: [0x24,0xec,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x24,0xec,0x99,0xae]
+
+# CHECK: vstrb.16 q5, [r3], #-25 @ encoding: [0x23,0xec,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x23,0xec,0x99,0xae]
+
+# CHECK: vstrb.16 q5, [r2, #-25] @ encoding: [0x02,0xed,0x99,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x02,0xed,0x99,0xae]
+
+# CHECK: vstrb.16 q5, [r2, #-64] @ encoding: [0x02,0xed,0xc0,0xae]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x02,0xed,0xc0,0xae]
+
+# CHECK: vldrb.u32 q0, [r0] @ encoding: [0x90,0xfd,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfd,0x00,0x0f]
+
+# CHECK: vldrb.u32 q1, [r0] @ encoding: [0x90,0xfd,0x00,0x2f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfd,0x00,0x2f]
+
+# CHECK: vldrb.u32 q0, [r7] @ encoding: [0x97,0xfd,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xfd,0x00,0x0f]
+
+# CHECK: vldrb.u32 q3, [r7] @ encoding: [0x97,0xfd,0x00,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xfd,0x00,0x6f]
+
+# CHECK: vldrb.u32 q0, [r4, #56] @ encoding: [0x94,0xfd,0x38,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xfd,0x38,0x0f]
+
+# CHECK: vldrb.u32 q4, [r4, #56] @ encoding: [0x94,0xfd,0x38,0x8f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xfd,0x38,0x8f]
+
+# CHECK: vldrb.u32 q0, [r2, #56] @ encoding: [0x92,0xfd,0x38,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x38,0x0f]
+
+# CHECK: vldrb.u32 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0x38,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xfd,0x38,0xaf]
+
+# CHECK: vldrb.u32 q5, [r4, #56]! @ encoding: [0xb4,0xfd,0x38,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xfd,0x38,0xaf]
+
+# CHECK: vldrb.u32 q5, [r4], #-25 @ encoding: [0x34,0xfc,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x34,0xfc,0x19,0xaf]
+
+# CHECK: vldrb.u32 q5, [r3], #-25 @ encoding: [0x33,0xfc,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x33,0xfc,0x19,0xaf]
+
+# CHECK: vldrb.u32 q5, [r6, #-25] @ encoding: [0x16,0xfd,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xfd,0x19,0xaf]
+
+# CHECK: vldrb.u32 q5, [r6, #-64] @ encoding: [0x16,0xfd,0x40,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xfd,0x40,0xaf]
+
+# CHECK: vldrb.s32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x00,0x0f]
+
+# CHECK: vldrb.s32 q1, [r0] @ encoding: [0x90,0xed,0x00,0x2f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x00,0x2f]
+
+# CHECK: vldrb.s32 q0, [r7] @ encoding: [0x97,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xed,0x00,0x0f]
+
+# CHECK: vldrb.s32 q3, [r7] @ encoding: [0x97,0xed,0x00,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x97,0xed,0x00,0x6f]
+
+# CHECK: vldrb.s32 q0, [r4, #56] @ encoding: [0x94,0xed,0x38,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x38,0x0f]
+
+# CHECK: vldrb.s32 q4, [r4, #56] @ encoding: [0x94,0xed,0x38,0x8f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x38,0x8f]
+
+# CHECK: vldrb.s32 q0, [r2, #56] @ encoding: [0x92,0xed,0x38,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xed,0x38,0x0f]
+
+# CHECK: vldrb.s32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x38,0xaf]
+
+# CHECK: vldrb.s32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x38,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x38,0xaf]
+
+# CHECK: vldrb.s32 q5, [r4], #-25 @ encoding: [0x34,0xec,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x34,0xec,0x19,0xaf]
+
+# CHECK: vldrb.s32 q5, [r3], #-25 @ encoding: [0x33,0xec,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x33,0xec,0x19,0xaf]
+
+# CHECK: vldrb.s32 q5, [r6, #-25] @ encoding: [0x16,0xed,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xed,0x19,0xaf]
+
+# CHECK: vldrb.s32 q5, [r6, #-64] @ encoding: [0x16,0xed,0x40,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x16,0xed,0x40,0xaf]
+
+# CHECK: vstrb.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x00,0x0f]
+
+# CHECK: vstrb.32 q1, [r0] @ encoding: [0x80,0xed,0x00,0x2f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x00,0x2f]
+
+# CHECK: vstrb.32 q0, [r7] @ encoding: [0x87,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x87,0xed,0x00,0x0f]
+
+# CHECK: vstrb.32 q3, [r7] @ encoding: [0x87,0xed,0x00,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x87,0xed,0x00,0x6f]
+
+# CHECK: vstrb.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x38,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x38,0x0f]
+
+# CHECK: vstrb.32 q4, [r4, #56] @ encoding: [0x84,0xed,0x38,0x8f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x38,0x8f]
+
+# CHECK: vstrb.32 q0, [r5, #56] @ encoding: [0x85,0xed,0x38,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x85,0xed,0x38,0x0f]
+
+# CHECK: vstrb.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x38,0xaf]
+
+# CHECK: vstrb.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x38,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x38,0xaf]
+
+# CHECK: vstrb.32 q5, [r4], #-25 @ encoding: [0x24,0xec,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x24,0xec,0x19,0xaf]
+
+# CHECK: vstrb.32 q5, [r3], #-25 @ encoding: [0x23,0xec,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x23,0xec,0x19,0xaf]
+
+# CHECK: vstrb.32 q5, [r2, #-25] @ encoding: [0x02,0xed,0x19,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x02,0xed,0x19,0xaf]
+
+# CHECK: vstrb.32 q5, [r2, #-64] @ encoding: [0x02,0xed,0x40,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x02,0xed,0x40,0xaf]
+
+# CHECK: vldrh.u16 q0, [r0] @ encoding: [0x90,0xed,0x80,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x80,0x1e]
+
+# CHECK: vldrh.u16 q1, [r0] @ encoding: [0x90,0xed,0x80,0x3e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x80,0x3e]
+
+# CHECK: vldrh.u16 q0, [r11] @ encoding: [0x9b,0xed,0x80,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9b,0xed,0x80,0x1e]
+
+# CHECK: vldrh.u16 q3, [r11] @ encoding: [0x9b,0xed,0x80,0x7e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9b,0xed,0x80,0x7e]
+
+# CHECK: vldrh.u16 q0, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x9c,0x1e]
+
+# CHECK: vldrh.u16 q4, [r4, #56] @ encoding: [0x94,0xed,0x9c,0x9e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x9c,0x9e]
+
+# CHECK: vldrh.u16 q0, [r8, #56] @ encoding: [0x98,0xed,0x9c,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xed,0x9c,0x1e]
+
+# CHECK: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x9c,0xbe]
+
+# CHECK: vldrh.u16 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x9c,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x9c,0xbe]
+
+# CHECK: vldrh.u16 q5, [r4], #-26 @ encoding: [0x34,0xec,0x8d,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x34,0xec,0x8d,0xbe]
+
+# CHECK: vldrh.u16 q5, [r10], #-26 @ encoding: [0x3a,0xec,0x8d,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x3a,0xec,0x8d,0xbe]
+
+# CHECK: vldrh.u16 q5, [sp, #-26] @ encoding: [0x1d,0xed,0x8d,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0x8d,0xbe]
+
+# CHECK: vldrh.u16 q5, [sp, #-64] @ encoding: [0x1d,0xed,0xa0,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0xa0,0xbe]
+
+# CHECK: vldrh.u16 q5, [sp, #-254] @ encoding: [0x1d,0xed,0xff,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0xff,0xbe]
+
+# CHECK: vldrh.u16 q5, [r10], #254 @ encoding: [0xba,0xec,0xff,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xba,0xec,0xff,0xbe]
+
+# CHECK: vstrh.16 q0, [r0] @ encoding: [0x80,0xed,0x80,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x80,0x1e]
+
+# CHECK: vstrh.16 q1, [r0] @ encoding: [0x80,0xed,0x80,0x3e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x80,0x3e]
+
+# CHECK: vstrh.16 q0, [r11] @ encoding: [0x8b,0xed,0x80,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8b,0xed,0x80,0x1e]
+
+# CHECK: vstrh.16 q3, [r11] @ encoding: [0x8b,0xed,0x80,0x7e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8b,0xed,0x80,0x7e]
+
+# CHECK: vstrh.16 q0, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x9c,0x1e]
+
+# CHECK: vstrh.16 q4, [r4, #56] @ encoding: [0x84,0xed,0x9c,0x9e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x9c,0x9e]
+
+# CHECK: vstrh.16 q0, [r8, #56] @ encoding: [0x88,0xed,0x9c,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x88,0xed,0x9c,0x1e]
+
+# CHECK: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x9c,0xbe]
+
+# CHECK: vstrh.16 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x9c,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x9c,0xbe]
+
+# CHECK: vstrh.16 q5, [r4], #-26 @ encoding: [0x24,0xec,0x8d,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x24,0xec,0x8d,0xbe]
+
+# CHECK: vstrh.16 q5, [r10], #-26 @ encoding: [0x2a,0xec,0x8d,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x2a,0xec,0x8d,0xbe]
+
+# CHECK: vstrh.16 q5, [sp, #-26] @ encoding: [0x0d,0xed,0x8d,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0x8d,0xbe]
+
+# CHECK: vstrh.16 q5, [sp, #-64] @ encoding: [0x0d,0xed,0xa0,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0xa0,0xbe]
+
+# CHECK: vstrh.16 q5, [sp, #-254] @ encoding: [0x0d,0xed,0xff,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0xff,0xbe]
+
+# CHECK: vstrh.16 q5, [r10], #254 @ encoding: [0xaa,0xec,0xff,0xbe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xaa,0xec,0xff,0xbe]
+
+# CHECK: vldrh.u32 q0, [r0] @ encoding: [0x98,0xfd,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xfd,0x00,0x0f]
+
+# CHECK: vldrh.u32 q1, [r0] @ encoding: [0x98,0xfd,0x00,0x2f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xfd,0x00,0x2f]
+
+# CHECK: vldrh.u32 q0, [r7] @ encoding: [0x9f,0xfd,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9f,0xfd,0x00,0x0f]
+
+# CHECK: vldrh.u32 q3, [r7] @ encoding: [0x9f,0xfd,0x00,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9f,0xfd,0x00,0x6f]
+
+# CHECK: vldrh.u32 q0, [r4, #56] @ encoding: [0x9c,0xfd,0x1c,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9c,0xfd,0x1c,0x0f]
+
+# CHECK: vldrh.u32 q4, [r4, #56] @ encoding: [0x9c,0xfd,0x1c,0x8f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9c,0xfd,0x1c,0x8f]
+
+# CHECK: vldrh.u32 q0, [r2, #56] @ encoding: [0x9a,0xfd,0x1c,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9a,0xfd,0x1c,0x0f]
+
+# CHECK: vldrh.u32 q5, [r4, #56]! @ encoding: [0xbc,0xfd,0x1c,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xbc,0xfd,0x1c,0xaf]
+
+# CHECK: vldrh.u32 q5, [r4, #56]! @ encoding: [0xbc,0xfd,0x1c,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xbc,0xfd,0x1c,0xaf]
+
+# CHECK: vldrh.u32 q5, [r4], #-26 @ encoding: [0x3c,0xfc,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x3c,0xfc,0x0d,0xaf]
+
+# CHECK: vldrh.u32 q5, [r3], #-26 @ encoding: [0x3b,0xfc,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x3b,0xfc,0x0d,0xaf]
+
+# CHECK: vldrh.u32 q5, [r6, #-26] @ encoding: [0x1e,0xfd,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1e,0xfd,0x0d,0xaf]
+
+# CHECK: vldrh.u32 q5, [r6, #-64] @ encoding: [0x1e,0xfd,0x20,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1e,0xfd,0x20,0xaf]
+
+# CHECK: vldrh.u32 q5, [r6, #-254] @ encoding: [0x1e,0xfd,0x7f,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1e,0xfd,0x7f,0xaf]
+
+# CHECK: vldrh.u32 q5, [r4, #254]! @ encoding: [0xbc,0xfd,0x7f,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xbc,0xfd,0x7f,0xaf]
+
+# CHECK: vldrh.s32 q0, [r0] @ encoding: [0x98,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xed,0x00,0x0f]
+
+# CHECK: vldrh.s32 q1, [r0] @ encoding: [0x98,0xed,0x00,0x2f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xed,0x00,0x2f]
+
+# CHECK: vldrh.s32 q0, [r7] @ encoding: [0x9f,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9f,0xed,0x00,0x0f]
+
+# CHECK: vldrh.s32 q3, [r7] @ encoding: [0x9f,0xed,0x00,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9f,0xed,0x00,0x6f]
+
+# CHECK: vldrh.s32 q0, [r4, #56] @ encoding: [0x9c,0xed,0x1c,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9c,0xed,0x1c,0x0f]
+
+# CHECK: vldrh.s32 q4, [r4, #56] @ encoding: [0x9c,0xed,0x1c,0x8f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9c,0xed,0x1c,0x8f]
+
+# CHECK: vldrh.s32 q0, [r2, #56] @ encoding: [0x9a,0xed,0x1c,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9a,0xed,0x1c,0x0f]
+
+# CHECK: vldrh.s32 q5, [r4, #56]! @ encoding: [0xbc,0xed,0x1c,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xbc,0xed,0x1c,0xaf]
+
+# CHECK: vldrh.s32 q5, [r4, #56]! @ encoding: [0xbc,0xed,0x1c,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xbc,0xed,0x1c,0xaf]
+
+# CHECK: vldrh.s32 q5, [r4], #-26 @ encoding: [0x3c,0xec,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x3c,0xec,0x0d,0xaf]
+
+# CHECK: vldrh.s32 q5, [r3], #-26 @ encoding: [0x3b,0xec,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x3b,0xec,0x0d,0xaf]
+
+# CHECK: vldrh.s32 q5, [r6, #-26] @ encoding: [0x1e,0xed,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1e,0xed,0x0d,0xaf]
+
+# CHECK: vldrh.s32 q5, [r6, #-64] @ encoding: [0x1e,0xed,0x20,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1e,0xed,0x20,0xaf]
+
+# CHECK: vldrh.s32 q5, [r6, #-254] @ encoding: [0x1e,0xed,0x7f,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1e,0xed,0x7f,0xaf]
+
+# CHECK: vldrh.s32 q5, [r4, #254]! @ encoding: [0xbc,0xed,0x7f,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xbc,0xed,0x7f,0xaf]
+
+# CHECK: vstrh.32 q0, [r0] @ encoding: [0x88,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x88,0xed,0x00,0x0f]
+
+# CHECK: vstrh.32 q1, [r0] @ encoding: [0x88,0xed,0x00,0x2f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x88,0xed,0x00,0x2f]
+
+# CHECK: vstrh.32 q0, [r7] @ encoding: [0x8f,0xed,0x00,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8f,0xed,0x00,0x0f]
+
+# CHECK: vstrh.32 q3, [r7] @ encoding: [0x8f,0xed,0x00,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8f,0xed,0x00,0x6f]
+
+# CHECK: vstrh.32 q0, [r4, #56] @ encoding: [0x8c,0xed,0x1c,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8c,0xed,0x1c,0x0f]
+
+# CHECK: vstrh.32 q4, [r4, #56] @ encoding: [0x8c,0xed,0x1c,0x8f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8c,0xed,0x1c,0x8f]
+
+# CHECK: vstrh.32 q0, [r5, #56] @ encoding: [0x8d,0xed,0x1c,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8d,0xed,0x1c,0x0f]
+
+# CHECK: vstrh.32 q5, [r4, #56]! @ encoding: [0xac,0xed,0x1c,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xac,0xed,0x1c,0xaf]
+
+# CHECK: vstrh.32 q5, [r4, #56]! @ encoding: [0xac,0xed,0x1c,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xac,0xed,0x1c,0xaf]
+
+# CHECK: vstrh.32 q5, [r4], #-26 @ encoding: [0x2c,0xec,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x2c,0xec,0x0d,0xaf]
+
+# CHECK: vstrh.32 q5, [r3], #-26 @ encoding: [0x2b,0xec,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x2b,0xec,0x0d,0xaf]
+
+# CHECK: vstrh.32 q5, [r2, #-26] @ encoding: [0x0a,0xed,0x0d,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0a,0xed,0x0d,0xaf]
+
+# CHECK: vstrh.32 q5, [r2, #-64] @ encoding: [0x0a,0xed,0x20,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0a,0xed,0x20,0xaf]
+
+# CHECK: vstrh.32 q5, [r2, #-254] @ encoding: [0x0a,0xed,0x7f,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0a,0xed,0x7f,0xaf]
+
+# CHECK: vstrh.32 q5, [r4, #254]! @ encoding: [0xac,0xed,0x7f,0xaf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xac,0xed,0x7f,0xaf]
+
+# CHECK: vldrw.u32 q0, [r0] @ encoding: [0x90,0xed,0x00,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x00,0x1f]
+
+# CHECK: vldrw.u32 q1, [r0] @ encoding: [0x90,0xed,0x00,0x3f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xed,0x00,0x3f]
+
+# CHECK: vldrw.u32 q0, [r11] @ encoding: [0x9b,0xed,0x00,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9b,0xed,0x00,0x1f]
+
+# CHECK: vldrw.u32 q3, [r11] @ encoding: [0x9b,0xed,0x00,0x7f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9b,0xed,0x00,0x7f]
+
+# CHECK: vldrw.u32 q0, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x0e,0x1f]
+
+# CHECK: vldrw.u32 q4, [r4, #56] @ encoding: [0x94,0xed,0x0e,0x9f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x94,0xed,0x0e,0x9f]
+
+# CHECK: vldrw.u32 q0, [r8, #56] @ encoding: [0x98,0xed,0x0e,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xed,0x0e,0x1f]
+
+# CHECK: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x0e,0xbf]
+
+# CHECK: vldrw.u32 q5, [r4, #56]! @ encoding: [0xb4,0xed,0x0e,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x0e,0xbf]
+
+# CHECK: vldrw.u32 q5, [r4], #-28 @ encoding: [0x34,0xec,0x07,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x34,0xec,0x07,0xbf]
+
+# CHECK: vldrw.u32 q5, [r10], #-28 @ encoding: [0x3a,0xec,0x07,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x3a,0xec,0x07,0xbf]
+
+# CHECK: vldrw.u32 q5, [sp, #-28] @ encoding: [0x1d,0xed,0x07,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0x07,0xbf]
+
+# CHECK: vldrw.u32 q5, [sp, #-64] @ encoding: [0x1d,0xed,0x10,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0x10,0xbf]
+
+# CHECK: vldrw.u32 q5, [sp, #-508] @ encoding: [0x1d,0xed,0x7f,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x1d,0xed,0x7f,0xbf]
+
+# CHECK: vldrw.u32 q5, [r4, #508]! @ encoding: [0xb4,0xed,0x7f,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb4,0xed,0x7f,0xbf]
+
+# ERROR: [[@LINE+2]]:2: warning: potentially undefined instruction encoding
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xbd,0xed,0x7f,0xbf]
+
+# CHECK: vstrw.32 q0, [r0] @ encoding: [0x80,0xed,0x00,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x00,0x1f]
+
+# CHECK: vstrw.32 q1, [r0] @ encoding: [0x80,0xed,0x00,0x3f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xed,0x00,0x3f]
+
+# CHECK: vstrw.32 q0, [r11] @ encoding: [0x8b,0xed,0x00,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8b,0xed,0x00,0x1f]
+
+# CHECK: vstrw.32 q3, [r11] @ encoding: [0x8b,0xed,0x00,0x7f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8b,0xed,0x00,0x7f]
+
+# CHECK: vstrw.32 q0, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x0e,0x1f]
+
+# CHECK: vstrw.32 q4, [r4, #56] @ encoding: [0x84,0xed,0x0e,0x9f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x84,0xed,0x0e,0x9f]
+
+# CHECK: vstrw.32 q0, [r8, #56] @ encoding: [0x88,0xed,0x0e,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x88,0xed,0x0e,0x1f]
+
+# CHECK: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x0e,0xbf]
+
+# CHECK: vstrw.32 q5, [r4, #56]! @ encoding: [0xa4,0xed,0x0e,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x0e,0xbf]
+
+# CHECK: vstrw.32 q5, [r4], #-28 @ encoding: [0x24,0xec,0x07,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x24,0xec,0x07,0xbf]
+
+# CHECK: vstrw.32 q5, [r10], #-28 @ encoding: [0x2a,0xec,0x07,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x2a,0xec,0x07,0xbf]
+
+# CHECK: vstrw.32 q5, [sp, #-28] @ encoding: [0x0d,0xed,0x07,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0x07,0xbf]
+
+# CHECK: vstrw.32 q5, [sp, #-64] @ encoding: [0x0d,0xed,0x10,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0x10,0xbf]
+
+# CHECK: vstrw.32 q5, [sp, #-508] @ encoding: [0x0d,0xed,0x7f,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x0d,0xed,0x7f,0xbf]
+
+# CHECK: vstrw.32 q5, [r4, #508]! @ encoding: [0xa4,0xed,0x7f,0xbf]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa4,0xed,0x7f,0xbf]
+
+# ERROR: [[@LINE+2]]:2: warning: potentially undefined instruction encoding
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xad,0xed,0x7f,0xbf]
+
+# CHECK: vldrb.u8 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x02,0x0e]
+
+# CHECK: vldrb.u8 q3, [r10, q1] @ encoding: [0x9a,0xfc,0x02,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9a,0xfc,0x02,0x6e]
+
+# CHECK: vldrb.u8 q3, [r0, q3] @ encoding: [0x90,0xfc,0x06,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x06,0x6e]
+
+# CHECK: vldrb.u16 q0, [r0, q1] @ encoding: [0x90,0xfc,0x82,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x82,0x0e]
+
+# CHECK: vldrb.u16 q3, [r9, q1] @ encoding: [0x99,0xfc,0x82,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x99,0xfc,0x82,0x6e]
+
+# CHECK: vldrb.u16 q3, [r0, q3] @ encoding: [0x90,0xfc,0x86,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x86,0x6e]
+
+# CHECK: vldrb.s16 q0, [r0, q1] @ encoding: [0x90,0xec,0x82,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x82,0x0e]
+
+# CHECK: vldrb.s16 q3, [sp, q1] @ encoding: [0x9d,0xec,0x82,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9d,0xec,0x82,0x6e]
+
+# CHECK: vldrb.s16 q3, [r0, q3] @ encoding: [0x90,0xec,0x86,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x86,0x6e]
+
+# CHECK: vldrb.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x02,0x0f]
+
+# CHECK: vldrb.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x02,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x02,0x6f]
+
+# CHECK: vldrb.u32 q3, [r0, q3] @ encoding: [0x90,0xfc,0x06,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x06,0x6f]
+
+# CHECK: vldrb.s32 q0, [r0, q1] @ encoding: [0x90,0xec,0x02,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x02,0x0f]
+
+# CHECK: vldrb.s32 q3, [r0, q1] @ encoding: [0x90,0xec,0x02,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x02,0x6f]
+
+# CHECK: vldrb.s32 q3, [r0, q3] @ encoding: [0x90,0xec,0x06,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x06,0x6f]
+
+# CHECK: vldrh.u16 q0, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x92,0x0e]
+
+# CHECK: vldrh.u16 q3, [r0, q1] @ encoding: [0x90,0xfc,0x92,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x92,0x6e]
+
+# CHECK: vldrh.u16 q3, [r0, q3] @ encoding: [0x90,0xfc,0x96,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x96,0x6e]
+
+# CHECK: vldrh.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x12,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x12,0x0f]
+
+# CHECK: vldrh.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x12,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x12,0x6f]
+
+# CHECK: vldrh.u32 q3, [r0, q3] @ encoding: [0x90,0xfc,0x16,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x16,0x6f]
+
+# CHECK: vldrh.s32 q0, [r0, q1] @ encoding: [0x90,0xec,0x12,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x12,0x0f]
+
+# CHECK: vldrh.s32 q3, [r0, q1] @ encoding: [0x90,0xec,0x12,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x12,0x6f]
+
+# CHECK: vldrh.s32 q3, [r0, q3] @ encoding: [0x90,0xec,0x16,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xec,0x16,0x6f]
+
+# CHECK: vldrh.u16 q0, [r0, q1, uxtw #1] @ encoding: [0x90,0xfc,0x93,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x93,0x0e]
+
+# CHECK: vldrh.u32 q3, [r8, q3, uxtw #1] @ encoding: [0x98,0xfc,0x17,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x98,0xfc,0x17,0x6f]
+
+# CHECK: vldrw.u32 q0, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x42,0x0f]
+
+# CHECK: vldrw.u32 q3, [r0, q1] @ encoding: [0x90,0xfc,0x42,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x42,0x6f]
+
+# CHECK: vldrw.u32 q3, [r0, q3] @ encoding: [0x90,0xfc,0x46,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x46,0x6f]
+
+# CHECK: vldrw.u32 q0, [r0, q1, uxtw #2] @ encoding: [0x90,0xfc,0x43,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0x43,0x0f]
+
+# CHECK: vldrw.u32 q0, [sp, q1, uxtw #2] @ encoding: [0x9d,0xfc,0x43,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9d,0xfc,0x43,0x0f]
+
+# CHECK: vldrd.u64 q0, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0xd2,0x0f]
+
+# CHECK: vldrd.u64 q3, [r0, q1] @ encoding: [0x90,0xfc,0xd2,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0xd2,0x6f]
+
+# CHECK: vldrd.u64 q3, [r0, q3] @ encoding: [0x90,0xfc,0xd6,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0xd6,0x6f]
+
+# CHECK: vldrd.u64 q0, [r0, q1, uxtw #3] @ encoding: [0x90,0xfc,0xd3,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x90,0xfc,0xd3,0x0f]
+
+# CHECK: vldrd.u64 q0, [sp, q1, uxtw #3] @ encoding: [0x9d,0xfc,0xd3,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9d,0xfc,0xd3,0x0f]
+
+# CHECK: vstrb.8 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x02,0x0e]
+
+# CHECK: vstrb.8 q3, [r10, q1] @ encoding: [0x8a,0xec,0x02,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8a,0xec,0x02,0x6e]
+
+# CHECK: vstrb.8 q3, [r0, q3] @ encoding: [0x80,0xec,0x06,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x06,0x6e]
+
+# CHECK: vstrb.16 q0, [r0, q1] @ encoding: [0x80,0xec,0x82,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x82,0x0e]
+
+# CHECK: vstrb.16 q3, [sp, q1] @ encoding: [0x8d,0xec,0x82,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8d,0xec,0x82,0x6e]
+
+# CHECK: vstrb.16 q3, [r0, q3] @ encoding: [0x80,0xec,0x86,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x86,0x6e]
+
+# CHECK: vstrb.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x02,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x02,0x0f]
+
+# CHECK: vstrb.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x02,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x02,0x6f]
+
+# CHECK: vstrb.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x06,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x06,0x6f]
+
+# CHECK: vstrh.16 q0, [r0, q1] @ encoding: [0x80,0xec,0x92,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x92,0x0e]
+
+# CHECK: vstrh.16 q3, [r0, q1] @ encoding: [0x80,0xec,0x92,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x92,0x6e]
+
+# CHECK: vstrh.16 q3, [r0, q3] @ encoding: [0x80,0xec,0x96,0x6e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x96,0x6e]
+
+# CHECK: vstrh.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x12,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x12,0x0f]
+
+# CHECK: vstrh.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x12,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x12,0x6f]
+
+# CHECK: vstrh.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x16,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x16,0x6f]
+
+# CHECK: vstrh.16 q0, [r0, q1, uxtw #1] @ encoding: [0x80,0xec,0x93,0x0e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x93,0x0e]
+
+# CHECK: vstrh.32 q3, [r8, q3, uxtw #1] @ encoding: [0x88,0xec,0x17,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x88,0xec,0x17,0x6f]
+
+# CHECK: vstrw.32 q0, [r0, q1] @ encoding: [0x80,0xec,0x42,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x42,0x0f]
+
+# CHECK: vstrw.32 q3, [r0, q1] @ encoding: [0x80,0xec,0x42,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x42,0x6f]
+
+# CHECK: vstrw.32 q3, [r0, q3] @ encoding: [0x80,0xec,0x46,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x46,0x6f]
+
+# CHECK: vstrw.32 q0, [r0, q1, uxtw #2] @ encoding: [0x80,0xec,0x43,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0x43,0x0f]
+
+# CHECK: vstrw.32 q0, [sp, q1, uxtw #2] @ encoding: [0x8d,0xec,0x43,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8d,0xec,0x43,0x0f]
+
+# CHECK: vstrd.64 q0, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0xd2,0x0f]
+
+# CHECK: vstrd.64 q3, [r0, q1] @ encoding: [0x80,0xec,0xd2,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0xd2,0x6f]
+
+# CHECK: vstrd.64 q3, [r0, q3] @ encoding: [0x80,0xec,0xd6,0x6f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0xd6,0x6f]
+
+# CHECK: vstrd.64 q0, [r0, q1, uxtw #3] @ encoding: [0x80,0xec,0xd3,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x80,0xec,0xd3,0x0f]
+
+# CHECK: vstrd.64 q0, [sp, q1, uxtw #3] @ encoding: [0x8d,0xec,0xd3,0x0f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8d,0xec,0xd3,0x0f]
+
+# CHECK: vldrw.u32 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x00,0x1e]
+
+# CHECK: vldrw.u32 q7, [q1] @ encoding: [0x92,0xfd,0x00,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x00,0xfe]
+
+# CHECK: vldrw.u32 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb2,0xfd,0x00,0xfe]
+
+# CHECK: vldrw.u32 q7, [q7] @ encoding: [0x9e,0xfd,0x00,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9e,0xfd,0x00,0xfe]
+
+# CHECK: vldrw.u32 q7, [q1, #4] @ encoding: [0x92,0xfd,0x01,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x01,0xfe]
+
+# CHECK: vldrw.u32 q7, [q1, #-4] @ encoding: [0x12,0xfd,0x01,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x12,0xfd,0x01,0xfe]
+
+# CHECK: vldrw.u32 q7, [q1, #508] @ encoding: [0x92,0xfd,0x7f,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x7f,0xfe]
+
+# CHECK: vldrw.u32 q7, [q1, #-508] @ encoding: [0x12,0xfd,0x7f,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x12,0xfd,0x7f,0xfe]
+
+# CHECK: vldrw.u32 q7, [q1, #264] @ encoding: [0x92,0xfd,0x42,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x42,0xfe]
+
+# CHECK: vstrw.32 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1e]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x00,0x1e]
+
+# CHECK: vstrw.32 q7, [q1] @ encoding: [0x82,0xfd,0x00,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x00,0xfe]
+
+# CHECK: vstrw.32 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa2,0xfd,0x00,0xfe]
+
+# CHECK: vstrw.32 q7, [q7] @ encoding: [0x8e,0xfd,0x00,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8e,0xfd,0x00,0xfe]
+
+# CHECK: vstrw.32 q7, [q1, #4] @ encoding: [0x82,0xfd,0x01,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x01,0xfe]
+
+# CHECK: vstrw.32 q7, [q1, #-4] @ encoding: [0x02,0xfd,0x01,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x02,0xfd,0x01,0xfe]
+
+# CHECK: vstrw.32 q7, [q1, #508] @ encoding: [0x82,0xfd,0x7f,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x7f,0xfe]
+
+# CHECK: vstrw.32 q7, [q1, #-508] @ encoding: [0x02,0xfd,0x7f,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x02,0xfd,0x7f,0xfe]
+
+# CHECK: vstrw.32 q7, [q1, #264]! @ encoding: [0xa2,0xfd,0x42,0xfe]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa2,0xfd,0x42,0xfe]
+
+# CHECK: vldrd.u64 q0, [q1] @ encoding: [0x92,0xfd,0x00,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x00,0x1f]
+
+# CHECK: vldrd.u64 q7, [q1] @ encoding: [0x92,0xfd,0x00,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x00,0xff]
+
+# CHECK: vldrd.u64 q7, [q1]! @ encoding: [0xb2,0xfd,0x00,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xb2,0xfd,0x00,0xff]
+
+# CHECK: vldrd.u64 q7, [q7] @ encoding: [0x9e,0xfd,0x00,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x9e,0xfd,0x00,0xff]
+
+# CHECK: vldrd.u64 q7, [q1, #8] @ encoding: [0x92,0xfd,0x01,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x01,0xff]
+
+# CHECK: vldrd.u64 q7, [q1, #-8] @ encoding: [0x12,0xfd,0x01,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x12,0xfd,0x01,0xff]
+
+# CHECK: vldrd.u64 q7, [q1, #1016] @ encoding: [0x92,0xfd,0x7f,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x7f,0xff]
+
+# CHECK: vldrd.u64 q7, [q1, #-1016] @ encoding: [0x12,0xfd,0x7f,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x12,0xfd,0x7f,0xff]
+
+# CHECK: vldrd.u64 q7, [q1, #264] @ encoding: [0x92,0xfd,0x21,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x21,0xff]
+
+# CHECK: vldrd.u64 q7, [q1, #624] @ encoding: [0x92,0xfd,0x4e,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x4e,0xff]
+
+# CHECK: vldrd.u64 q7, [q1, #264] @ encoding: [0x92,0xfd,0x21,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x92,0xfd,0x21,0xff]
+
+# CHECK: vstrd.64 q0, [q1] @ encoding: [0x82,0xfd,0x00,0x1f]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x00,0x1f]
+
+# CHECK: vstrd.64 q7, [q1] @ encoding: [0x82,0xfd,0x00,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x00,0xff]
+
+# CHECK: vstrd.64 q7, [q1]! @ encoding: [0xa2,0xfd,0x00,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0xa2,0xfd,0x00,0xff]
+
+# CHECK: vstrd.64 q7, [q7] @ encoding: [0x8e,0xfd,0x00,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x8e,0xfd,0x00,0xff]
+
+# CHECK: vstrd.64 q7, [q1, #8] @ encoding: [0x82,0xfd,0x01,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x01,0xff]
+
+# CHECK: vstrd.64 q7, [q1, #-8]! @ encoding: [0x22,0xfd,0x01,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x22,0xfd,0x01,0xff]
+
+# CHECK: vstrd.64 q7, [q1, #1016] @ encoding: [0x82,0xfd,0x7f,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x7f,0xff]
+
+# CHECK: vstrd.64 q7, [q1, #-1016] @ encoding: [0x02,0xfd,0x7f,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x02,0xfd,0x7f,0xff]
+
+# CHECK: vstrd.64 q7, [q1, #264] @ encoding: [0x82,0xfd,0x21,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x21,0xff]
+
+# CHECK: vstrd.64 q7, [q1, #624] @ encoding: [0x82,0xfd,0x4e,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x4e,0xff]
+
+# CHECK: vstrd.64 q7, [q1, #264] @ encoding: [0x82,0xfd,0x21,0xff]
+# CHECK-NOMVE: [[@LINE+1]]:2: warning: invalid instruction encoding
+[0x82,0xfd,0x21,0xff]
+
+# CHECK: vldrb.u8 q1, [r2] @ encoding: [0x92,0xed,0x00,0x3e]
+[0x92,0xed,0x00,0x3e]
+
+# CHECK: vldrb.u8 q1, [r2, #-0] @ encoding: [0x12,0xed,0x00,0x3e]
+[0x12,0xed,0x00,0x3e]
+
+# CHECK: vldrb.u16 q1, [r2] @ encoding: [0x92,0xfd,0x80,0x2e]
+[0x92,0xfd,0x80,0x2e]
+
+# CHECK: vldrb.u16 q1, [r2, #-0] @ encoding: [0x12,0xfd,0x80,0x2e]
+[0x12,0xfd,0x80,0x2e]
+
+# CHECK: vldrb.s16 q1, [r2] @ encoding: [0x92,0xed,0x80,0x2e]
+[0x92,0xed,0x80,0x2e]
+
+# CHECK: vldrb.s16 q1, [r2, #-0] @ encoding: [0x12,0xed,0x80,0x2e]
+[0x12,0xed,0x80,0x2e]
+
+# CHECK: vldrb.u32 q1, [r2] @ encoding: [0x92,0xfd,0x00,0x2f]
+[0x92,0xfd,0x00,0x2f]
+
+# CHECK: vldrb.u32 q1, [r2, #-0] @ encoding: [0x12,0xfd,0x00,0x2f]
+[0x12,0xfd,0x00,0x2f]
+
+# CHECK: vldrb.s32 q1, [r2] @ encoding: [0x92,0xed,0x00,0x2f]
+[0x92,0xed,0x00,0x2f]
+
+# CHECK: vldrb.s32 q1, [r2, #-0] @ encoding: [0x12,0xed,0x00,0x2f]
+[0x12,0xed,0x00,0x2f]
+
+# CHECK: vldrh.u16 q1, [r2] @ encoding: [0x92,0xed,0x80,0x3e]
+[0x92,0xed,0x80,0x3e]
+
+# CHECK: vldrh.u16 q1, [r2, #-0] @ encoding: [0x12,0xed,0x80,0x3e]
+[0x12,0xed,0x80,0x3e]
+
+# CHECK: vldrh.u32 q1, [r2] @ encoding: [0x9a,0xfd,0x00,0x2f]
+[0x9a,0xfd,0x00,0x2f]
+
+# CHECK: vldrh.u32 q1, [r2, #-0] @ encoding: [0x1a,0xfd,0x00,0x2f]
+[0x1a,0xfd,0x00,0x2f]
+
+# CHECK: vldrh.s32 q1, [r2] @ encoding: [0x9a,0xed,0x00,0x2f]
+[0x9a,0xed,0x00,0x2f]
+
+# CHECK: vldrh.s32 q1, [r2, #-0] @ encoding: [0x1a,0xed,0x00,0x2f]
+[0x1a,0xed,0x00,0x2f]
+
+# CHECK: vldrw.u32 q1, [r2] @ encoding: [0x92,0xed,0x00,0x3f]
+[0x92,0xed,0x00,0x3f]
+
+# CHECK: vldrw.u32 q1, [r2, #-0] @ encoding: [0x12,0xed,0x00,0x3f]
+[0x12,0xed,0x00,0x3f]
+
+# CHECK: vldrw.u32 q1, [q2] @ encoding: [0x94,0xfd,0x00,0x3e]
+[0x94,0xfd,0x00,0x3e]
+
+# CHECK: vldrw.u32 q1, [q2, #-0] @ encoding: [0x14,0xfd,0x00,0x3e]
+[0x14,0xfd,0x00,0x3e]
+
+# CHECK: vldrd.u64 q1, [q2] @ encoding: [0x94,0xfd,0x00,0x3f]
+[0x94,0xfd,0x00,0x3f]
+
+# CHECK: vldrd.u64 q1, [q2, #-0] @ encoding: [0x14,0xfd,0x00,0x3f]
+[0x14,0xfd,0x00,0x3f]
+
+# CHECK: vstrb.8 q1, [r2] @ encoding: [0x82,0xed,0x00,0x3e]
+[0x82,0xed,0x00,0x3e]
+
+# CHECK: vstrb.8 q1, [r2, #-0] @ encoding: [0x02,0xed,0x00,0x3e]
+[0x02,0xed,0x00,0x3e]
+
+# CHECK: vstrb.16 q1, [r2] @ encoding: [0x82,0xed,0x80,0x2e]
+[0x82,0xed,0x80,0x2e]
+
+# CHECK: vstrb.16 q1, [r2, #-0] @ encoding: [0x02,0xed,0x80,0x2e]
+[0x02,0xed,0x80,0x2e]
+
+# CHECK: vstrb.32 q1, [r2] @ encoding: [0x82,0xed,0x00,0x2f]
+[0x82,0xed,0x00,0x2f]
+
+# CHECK: vstrb.32 q1, [r2, #-0] @ encoding: [0x02,0xed,0x00,0x2f]
+[0x02,0xed,0x00,0x2f]
+
+# CHECK: vstrh.16 q1, [r2] @ encoding: [0x82,0xed,0x80,0x3e]
+[0x82,0xed,0x80,0x3e]
+
+# CHECK: vstrh.16 q1, [r2, #-0] @ encoding: [0x02,0xed,0x80,0x3e]
+[0x02,0xed,0x80,0x3e]
+
+# CHECK: vstrh.32 q1, [r2] @ encoding: [0x8a,0xed,0x00,0x2f]
+[0x8a,0xed,0x00,0x2f]
+
+# CHECK: vstrh.32 q1, [r2, #-0] @ encoding: [0x0a,0xed,0x00,0x2f]
+[0x0a,0xed,0x00,0x2f]
+
+# CHECK: vstrw.32 q1, [r2] @ encoding: [0x82,0xed,0x00,0x3f]
+[0x82,0xed,0x00,0x3f]
+
+# CHECK: vstrw.32 q1, [r2, #-0] @ encoding: [0x02,0xed,0x00,0x3f]
+[0x02,0xed,0x00,0x3f]
+
+# CHECK: vstrw.32 q1, [q2] @ encoding: [0x84,0xfd,0x00,0x3e]
+[0x84,0xfd,0x00,0x3e]
+
+# CHECK: vstrw.32 q1, [q2, #-0] @ encoding: [0x04,0xfd,0x00,0x3e]
+[0x04,0xfd,0x00,0x3e]
+
+# CHECK: vstrd.64 q1, [q2] @ encoding: [0x84,0xfd,0x00,0x3f]
+[0x84,0xfd,0x00,0x3f]
+
+# CHECK: vstrd.64 q1, [q2, #-0] @ encoding: [0x04,0xfd,0x00,0x3f]
+[0x04,0xfd,0x00,0x3f]
+